Whether newsrooms are paying enough attention or not, however, many governments are adopting AI strategies, policies and laws. These regulatory attempts rarely directly address journalism and vary dramatically in their frameworks, enforcement capacity and level of international coordination, but they will almost certainly have an impact on journalism and the digital information environment. To understand exactly how regulation is, or could be, impacting journalism, we reviewed 188 national and regional AI strategies, laws and policies that collectively cover more than 99 countries. 

The analysis is structured around the seven regions of the world, as identified by the World Bank — North America, Latin America and the Caribbean, Europe and Central Asia, the Middle East and North Africa, Sub-Saharan Africa, East Asia and the Pacific, and South Asia — and around how regulatory activity in each of these regions addresses seven policy components that, based on the research of CNTI and others, are particularly relevant to and likely to impact journalism:

Explore the proposals

CNTI reviewed 188 national and regional AI strategies, laws and policies around the world to determine how they impact journalism. We focused our analysis around how regulatory activity addresses seven policy components that are particularly likely to impact journalism. Use the below map to explore each of the policy documents we reviewed and to see what journalism-related topics they include. Use the column on the left to see which documents mention a specific topic, or click on a country to see the list of its documents and the topics covered in each. 

Key findings: The landscape

The governance of AI is increasingly complex and varies dramatically by country and industry. 

The broad range of governance instruments in play include legally-binding regulation at every level of government, corporate policies, industry standards and many other types of documents. AI applications are being introduced across sectors: some of them, such as banking and health, are already heavily regulated, while others are much more lightly regulated. Even among national-level legislation, there is tremendous diversity in scope: some bills are sweeping and comprehensive, while others comprise small edits to existing laws or address AI in narrowly-defined use cases like online shopping, mammograms or road accidents. Others create new agencies, committees or councils to operationalize vaguely laid out principles. The law-making process varies by country, as do the stakeholders involved. There is also variability in countries’ power to impact technology companies. In some smaller markets, companies may consider pulling out rather than comply with regulations, especially if those regulations are burdensome or do not align with the requirements of larger markets. Furthermore, some countries — the United States among them — have dozens or even hundreds of bills in the pipeline, making it impossible to conduct an exhaustive review. While we offer numbers wherever possible to give a sense of scale and proportion, our research examines the breadth of policies, strategies and laws rather than providing a full audit.

As with all policy proposals, it is important to consider the AI documents we reviewed in their individual country contexts and in relation to pre-existing laws in those countries and regions. Even if journalism or data privacy is not explicitly mentioned in an AI document, for example, existing protections for these rights may supersede new AI proposals. It is also important for policymakers to coordinate with experts to determine whether their AI policy proposals are technologically feasible as they consider replicating the documents reviewed in this paper or originating new AI policy proposals. 

Policies don’t have to mention journalism by name to impact it. 

Twenty of the 188 documents explicitly mentioned “journalist,” “journalism,” “news,” “media” or “news media.”

When these terms do appear, they range from passing references to the field to laws whose primary concern is the practice of journalism. This by no means is a suggestion that more documents should directly name journalism. Calling out the journalism field in policy documents frequently has its own pitfalls: Once governments define “journalism” or “news,” those definitions can be weaponized against the news media. Instead, it is important that policymakers are aware of and think through potential impacts of directly naming or not naming journalism and news. 

• Four documents (from Bahrain, Chile, Costa Rica and Kenya) identify journalists or media workers as stakeholders, but go no further. Three others (from the African Union, Panama and Serbia) specifically identify the field of journalism as an important audience for educational materials and reskilling. 
• Four documents (from Algeria, Egypt, Lesotho and Sri Lanka) emphasize the importance of news media as a communications channel for public awareness campaigns, with varying assumptions about editorial independence from government. 
• One proposed law (from Ecuador) attempts to intervene in polarization and fragmentation by requiring providers of content recommendation algorithms to include content from a broad range of media outlets.
• Five different bills and laws exempt journalism from specific provisions: China’s Interim Measures on Generative AI say that other journalism laws supersede it; three U.S. bills would exempt news media from specific restrictions on deepfakes in the context of both reporting and advertisements (Illinois, New Hampshire and South Dakota); and Brazil’s proposed law grants exemptions to some forms of copyright violation for journalistic or research purposes..
• Two proposed laws — both from the U.S., one in New Jersey and one in New York — would specifically regulate the use of AI in journalism.

Transparency and data protection are the two topic areas (among the seven studied) that come up most frequently.

Of the 188 AI strategies, laws and policies we reviewed, 124 addressed transparency and accountability; 107 addressed data protection and privacy; 92 addressed algorithmic discrimination and bias; 76 addressed public information and awareness about AI; 64 addressed manipulated or synthetic content; 49 addressed intellectual property and copyright; and 19 addressed freedom of speech and expression. Each document could be counted in multiple categories. In all seven regions, either transparency or data protection was the most common topic. In every region, more than half of the documents we reviewed addressed transparency and accountability. Likewise, in every region, more than half of the documents we reviewed addressed privacy and data protection, with the exception of North America, which only featured the topic in eight of the 29 documents we examined. The emphasis on transparency likely responds to a key challenge for accountability: the opaqueness of many AI systems leaves even their creators without a full understanding of how they work, let alone policymakers. Meanwhile, the salience of data privacy is to be expected, since it has been a key issue in technology governance over the last ten years. At least four out of five people around the world are protected by a national-level privacy law. 

Freedom of speech and expression come up least often.

While policymakers regularly express concern about AI’s impacts on the information environment, references to freedom of speech and expression are infrequent across regions. In fact, these topics did not appear in any documents we reviewed from either the Middle East and North Africa or Sub-Saharan Africa. Moreover, the documents we reviewed addressed two very different concerns: Some, like Malaysia, addressed the possibility that using AI would threaten fundamental human freedoms, while others, like Venezuela, addressed the possibility that regulating AI would threaten human freedoms. The latter was more common in the United States than elsewhere, but both issues came up across regions. The EU’s AI Act highlights both concerns, emphasizing that AI systems can violate fundamental freedoms but also that labeling obligations do not restrict free speech. One resolution, New Jersey’s AR 158, instead focuses on freedom of speech about AI by urging technology companies to embrace stronger whistleblower protections. We note that the absence of explicit language about freedom of speech or expression does not necessarily indicate an absence of concern; in many countries, these freedoms are guaranteed in foundational legal documents, like constitutions, that supersede all other laws or policies. The certainty around the upholding of those foundational documents, however, cannot be guaranteed.

Key findings: The topics

Freedom of speech and expression

Out of the 188 documents we reviewed, 20 address freedom of speech and expression directly.

CNTI’s analysis finds that when freedom of speech and expression are recognized, it generally has positive implications for journalism. This demonstrates an understanding of the importance of this freedom and an acknowledgement of how AI can impact it. Because it is often only included in preambles or as a guiding principle, however, these policies often do not specify how they are going to protect these rights. Examining the range of existing levels of press freedom, both across and within regions, is foundational to understanding how these policies may be used.

Of the few concrete policies about freedom of speech and expression, most are likely to benefit journalism. Several countries ban AI systems that do not respect freedom of expression, but they do not specify how or what uses of AI systems would fail to respect fundamental freedom. For example, Argentina’s Bill 2573-S-2024 prohibits the use of AI “which violates fundamental human rights such as privacy, freedom of expression, equality or human dignity.” Other countries recognize trade-offs: In order to ensure that algorithmic content recommendation does not violate freedom of expression, Ecuador’s Organic Law for the Regulation & Promotion of AI in Ecuador requires clear terms and conditions, human supervision, accountability reports and an appeals process. These provisions would make it more difficult for journalists to be subject to censorship at the hands of an overly conservative algorithm. 

Manipulated or synthetic content

Of the 188 documents we reviewed, 64 address manipulated or synthetic content directly. At least two documents from each region contain provisions about this issue.

CNTI’s analysis finds that attempts to prevent the spread of false information and highlight the provenance of information are broadly positive for journalism and the information space — but these efforts must be scrutinized to ensure they do not infringe on freedom of expression. 

For example, some proposals may protect journalists from technology-facilitated gender-based violence and other forms of AI-generated harassment. When a journalist is targeted with a deepfake, it can push them into silence, decrease their credibility and even drive them to leave the field. By banning explicit deepfakes, the United States’ Take It Down Act, Mexico’s Ley Olimpia and Bahrain’s draft AI Regulation Law would prohibit such attacks on journalists. 

However, some of these proposals may end up penalizing journalists. For example, the United States’ 2025 Take It Down Act prohibits the nonconsensual publication of both authentic and AI-generated intimate images and requires online platforms to remove such images. It does not, however, include safeguards against bad-faith or fraudulent takedown requests, which some advocates say could be used to wrongly censor journalism. 

Moreover, some provisions could potentially be used by the government to target journalists for reporting it does not like. Bahrain’s draft AI Regulation Law would prohibit using AI to “upload or install personal images that damage an individual’s reputation or dignity; [… or …] modify, edit, or tamper with textual, audio, or visual content related to individuals without their explicit consent.” These provisions are sufficiently vague that they could potentially be weaponized against journalists for innocuous changes, like editing the levels of an audio recording to make it clearer.

Finally, some proposals could impact journalists’ ability to use AI to protect their sources. The Dominican Republic’s proposed Bill 563, for example, would ban the use of deepfakes to alter videos and punish violators with prison terms and hefty fines. Because it does not include an exception for journalism, even when it is labeled, this could prevent journalists from using AI for legitimate purposes, such as to create an avatar of a source who wishes to remain anonymous for their safety.

Many laws also treat synthetic audio, images and video differently than synthetic text, increasing complexity in this space.

Algorithmic discrimination and bias

Out of the 188 documents we reviewed, 92 address algorithmic discrimination and bias directly. 

Some laws would limit the tools that news organizations can use for decision-making, but these laws would not uniquely impact this sector — they typically focus on the use of AI tools for hiring and other consequential decisions, and would apply universally. News organizations would simply have to comply with these regulations.

However, policies that focus on bias in content recommendation systems could potentially have strong impacts on the reach of journalistic content. Content recommendation systems impact the content people see online, especially on social media platforms. Some news organizations also use them to suggest content to audiences. If social media platforms are required to recommend a diversity of sources and opinions, this would impact the reach of journalistic content, but it’s hard to tell if it will extend or limit that reach. Some AI policies — particularly in the EU — would limit the recommender tools used by social media platforms, but not by news organizations themselves, making the impacts unpredictable. Elsewhere, such as in Ecuador, provisions would require news organizations to scrutinize their personalization or content delivery tools, such as AI-powered, customizable homepages.

Other documents, such as those in Bangladesh and Lesotho, would push journalists and newsrooms to be more cognizant of biases in tools they use to analyze data or create content, especially if these provisions require representative or diverse training data. Because journalists already prioritize objectivity, such awareness falls under good journalistic practice and is unlikely to penalize journalists.

Looking beyond journalistic uses of AI, some proposals create registries and audits, which may facilitate journalists’ ability to conduct accountability reporting on AI more broadly. For example, Lesotho’s Draft Artificial Intelligence Policy and Implementation Plan calls on the country’s AI regulator and policymakers to develop “bias mitigation programs,” by mandating bias audits, providing open-source tools for bias mitigation and detection, and training developers and other stakeholders on bias prevention. This could grant journalists greater insight into how these technologies work, thus improving their reporting capability and increasing public accountability. It could also make it easier for them to select appropriate tools for their professional use.

Intellectual property and copyright

Out of the 188 documents we reviewed, 49 address intellectual property and copyright directly. At least three documents from each region included this policy component.

Copyright and intellectual property regulations are particularly complicated because different countries’ laws are incompatible with each other, and few have reckoned with major shifts in distribution made possible by the internet. 

We found that some bills and laws require licensing and compensation for the use of copyrighted material to train AI models. Colombia’s Bill 293 states that beyond an exception for scientific uses, developers cannot use copyrighted content to train AI without prior and explicit consent. On the other hand, the Digital Single Market Directive of the EU AI Act, for example, does allow copyright holders to ‘opt-out’ of their content being used as training data for commercial purposes. The bill in Colombia also gives copyright collectives the right to authorize, prohibit, or restrict the use of works under their management, and they can demand just and equitable remuneration in order to license that use. These legal provisions will likely have positive financial impacts for journalism producers if they can be enforced.

At the same time, a number of documents mention the importance of protecting existing copyright and intellectual property laws without necessarily specifying how AI systems — particularly those using data in training models — fit into these legal foundations, let alone offering a comprehensive assessment of the value of these interactions with digital content. 

For example, China’s generative AI regulation stipulates that both deployers and developers need to comply with intellectual property laws in the country, but it does not specify how AI systems and AI-generated content fit into those laws. 

Japan, on the other hand, takes one of the most permissive stances, allowing the use of copyrighted works for AI training regardless of purpose, so long as it does not unreasonably prejudice the rights-holder’s interests. These divergent approaches raise the issue of interoperability when models are trained globally and deployed across borders. 

The matter of intellectual property and AI is far from settled, though. In November 2025, a court in the United Kingdom found that AI models are subject to copyright infringement claims; however, the lawsuit did not answer the question of whether using copyrighted materials to train AI models falls within the U.K.’s “fair dealing” provisions. Courts in the United States have taken differing views, with two judges in California ruling that AI companies’ use of copyrighted materials to train LLMs constitutes “fair use” because the work is “transformative,” albeit with caveats and for different reasons, while a judge in Delaware ruled that a different AI company’s use of copyrighted materials was not “transformative” and, thus, did not meet “fair use.” This area will continue to evolve as ongoing lawsuits determine whether copyrighted material can be used to train AI models around the world, including in the U.S., India and Japan. 

What is becoming clear is the need to holistically address the value of the various kinds of uses of and interactions with digital content in building AI models and beyond. Current policy approaches tend to dilate between narrow licensing regimes and permissive exceptions, but few adequately balance creators’ rights, developers’ needs and the public interest, let alone the vast array of content itself. 

Transparency and accountability

Out of the 188 documents we reviewed, 124 address transparency and accountability directly. At least three documents from each region included this policy component.

CNTI’s analysis finds that their impact on journalism will likely vary depending on whether transparency obligations fall more heavily on developers or deployers. If they fall on developers, such as in Malaysia, it will be easier for journalists to assess whether third-party tools are valuable to their work; if they fall on deployers, like in this Argentine proposal, journalists, as deployers, would be held responsible for choices made by third-party companies which could lead to greater attentiveness among journalists, and, if not, to unanticipated lawsuits. Several documents, however, fall somewhere in between. The United Arab Emirates’ AI Ethics Guide, for example, states that “accountability for the outcomes of an AI system lies not with the system itself but is apportioned between those who design, develop and deploy it,” meaning journalists could be held accountable at any stage of the process.

We also found that, in general, transparency requirements can make it easier for journalists to report across a range of topics and sectors. For example, California’s AI Transparency Act requires transparency on the data used to train AI models, which would increase public understanding of the tools and, thus, allow journalists to better assess these models.

Some proposals would also require news organizations to label AI-generated content to ensure the public recognizes when it is being used. New York’s proposed Senate Bill S6748, for example, would require publications to “conspicuously” identify at the top of a page or webpage when AI is used to either partially or wholly create an article, image, video or other piece of content. Such requirements could potentially be a form of “compelled speech” or could reduce trust in the content. 

Data protection and privacy 

Out of the 188 documents we reviewed, 107 address data protection and privacy directly.

AI systems and tools are trained on vast quantities of data, and many countries have comprehensive privacy legislation in place that will interact with AI regulation in complex ways.

AI legislation, policies and strategies that address data protection and privacy typically do not consider the specific needs of journalists, who require access to sensitive data for investigative reporting. Bangladesh’s National Artificial Intelligence Policy, for example, states that “personal data usage will require valid consent, notice, and the option to revoke.” Requiring journalists to receive consent before using personal details, however, could inhibit investigative reporting on corruption, human rights abuses and more, given that the politicians, business leaders and others in question would likely not grant consent for their data to be used.

On the other hand, CNTI’s analysis finds that proposals and laws that ban the use of AI for surveillance, as some countries have contemplated, would likely improve the safety of journalists and their sources. For example, the EU AI Act bans AI systems that “create or expand facial recognition databases through the untargeted scraping of facial images from the internet or CCTV footage.” While this represents a step in the right direction, the law does have exemptions for national security. Some countries have taken advantage of this; Hungary, for example, allowed the police to use biometric surveillance to identify participants in LGBTQIA+ public events. 

Public information and awareness

Out of the 188 documents we reviewed, 76 address this issue directly. At least two documents from each region directly touch on this issue.

Many provisions for broad public awareness include journalists among their audiences, providing up-to-date sociotechnical knowledge that could inform stronger reporting. For example, Serbia’s AI strategy includes plans to organize seminars on AI, information security and big data specifically for journalists. Other policies offer similar information to all adults. 

Documents that recognize the importance of journalism as a vehicle for public awareness are varied in their strategies. Lesotho’s draft policy encourages stakeholders to do outreach with a diverse spectrum of media outlets, acknowledging the importance of news without raising concerns about press independence. On the other hand, Egypt’s strategy expressly calls on the media to share “positive news of AI,” perhaps suggesting a bid to influence coverage. No document we reviewed calls for governments to spend advertising dollars to place public awareness campaigns in news media.

Recommendations

The inclusion or exclusion of these seven topics does not necessarily make a proposal “bad” or “good” overall or for journalism and the digital information space. There are many important considerations for AI policy proposals, of which journalism is just one. It is one, though, that CNTI finds incredibly important to functioning societies.

If an independent, diverse news media and open internet are not protected in the AI era, these new regulations could potentially criminalize journalism, threaten news business models, contribute to information disorder and prevent the public from accessing a diversity of fact-based news. The solution is not to call out journalism by name in every AI proposal, as that can have unintended consequences for the field, but it is essential that policymakers see journalists as important stakeholders in AI discussions and consider journalism’s viability as they develop future proposals. Likewise, it is important that news organizations and journalists see themselves as key stakeholders and thoroughly engage in thinking through and discussing the future of AI regulation.

The analysis surfaces a few key areas in need of specific policy attention: 

1. 1. In AI proposals that address manipulated content, it is important that policymakers work towards methods that protect certain journalistic uses in ways that do not enable government censorship or determination of who is or is not a journalist. This is far from an easy task and may mean the best path is no legal policy at all. Either way, it is critical to fully think through. Legislative efforts to prevent the spread of disinformation can sometimes have unintended consequences for journalism. For example, legislation that regulates the provision and sharing of manipulated content but does not include exceptions for journalism can lead to journalists being targeted for allegedly “spreading disinformation” for reporting on the existence of false information, such as a deepfake of an elected official, for using AI-manipulated content (like AI-generated avatars or voice-altered audio) to protect a source, or for covering information that those in power do not want covered. While not legally feasible in the U.S., carve outs for journalists in other countries could protect an independent news media, as well as a source’s right to privacy and willingness to share. But such carve outs, if not crafted extremely carefully, could also easily lead to greater censorship and criminalization. This takes coordinated, thoughtful deliberation. Alongside these discussions, journalists and news organizations should determine how best to convey their use of AI to the public, whether through labeling, watermarking or something else. 

2. 2. Bias audits and transparency measures are best implemented before a tool is deployed. To date, bias audits and transparency measures have often been reactions to issues that arise after deployment rather than components required prior to public launch. While not all issues can be identified in advance, more can be done before a tool is deployed, especially as these systems and our understanding of them matures. Policy can help lay out foundational requirements for audits, as well as for transparency, around how they are conducted and how issues are addressed, particularly for technology companies whose algorithms can have a social impact (i.e. if the algorithm uses personal data and makes or influences decisions that may have a significant impact on society or an individual). 

3. 3. Policymakers should ensure that AI working groups include journalism producers, product teams and engineers, alongside AI technologists, researchers, civil society and other relevant stakeholders. Policymakers face the challenge of attempting to legislate a technology that is constantly evolving and has wide-reaching consequences. As such, it is important that they meet with diverse stakeholders to ensure that they properly understand the technology they are trying to govern, as well as the consequences that both the technology and potential legislation could have on society more broadly. It is also critical that all at the table, including journalism leaders, are fully read-in and share the goal of collaboration, as such cooperation leads to the most optimal societal solutions. Expanded regional and international collaboration would also be valuable as AI regulation is a global challenge with cross-border implications. 

Acknowledgements

The authors would like to thank Monica Attard, Charlie Beckett, Niamh Burns, Claudia Del Pozo, Mohamed Farahat, Megan Gray, Assane Gueye, Jhalak Kakkar, Ashkhen Kazaryan, Tanit Koch, Prabhat Mishra, Amy Mitchell and Daniela Rojas for their thoughtful feedback on this report. 

The authors also extend their gratitude to Greta Alquist for editing this report, Jonathon Berlin and Ryan Marx for designing the graphics and maps, Kurt Cunningham for creating the web design and the team at CETRA for their translation work. 

The post Journalism’s New Frontier: An Analysis of Global AI Policy Proposals and Their Impacts on Journalism appeared first on Center for News, Technology & Innovation.

The United States has been experiencing an incredible amount of activity related to AI regulation. Much of that occurred in 2025 at the state level, as the likelihood of federal-level regulation largely evaporated.  Between January and June 2025,  over 1,000 bills — nearly all state-level — entered the legislative process. A massive consideration that loomed over state legislation was the U.S. Congress’s proposal, as part of the 2025 budget reconciliation process, to implement a 10-year moratorium on certain state-level regulation of AI. This moratorium was removed from the legislation prior to both chambers of Congress passing the bill in early July. (While outside this report’s timeline, it is worth noting that similar language reappeared in America’s AI Action Plan, released in July 2025, which is not a law, so does not hold legal authority, but sets the United States’s vision for global AI leadership and innovation.) 

At the national level, the current presidential administration has pushed an innovation-first approach to AI (with a few notable exceptions like the Take It Down Act in Congress). This differs from the last administration, which emphasized oversight and harm-based approaches to AI regulation. The shift in the federal approach to AI has resulted in increased state efforts to regulate the technology in the wake of the 2024 election; many state legislatures have ramped up legislative efforts to regulate AI, with a focus on balancing calls for innovation with the need to address harm. At the state level, some language repeats nearly verbatim across bills, particularly on the topics of deepfakes (especially in regard to elections), algorithmic bias and AI transparency. It is important to note that many of the approaches target “regulated industries” — like banking, healthcare or insurance — and as such would not apply directly to news and journalism.

In addition to the 50 U.S. states, territories of the United States are also developing proposals to regulate AI, like Guam with Bill No. 64-39 and Puerto Rico with SB 0068; these proposals cover early-stage regulation like creating task forces and commissions to study AI and formulate policies. Our analysis includes a variety of efforts at the national, state and territory level to demonstrate the breadth of approaches in the United States.

The increased activity across U.S. states has, at times, been met with legal pushback. For example, California’s AB 2839, which was intended to protect against deceptive deepfake content during election campaigns, was found to include unconstitutional infringements on freedom of speech and expression (e.g., satire and parody content). Legal challenges may continue across U.S. states as many are proposing legislation that is not in full alignment with recent Executive Orders. 

In Canada, activity initially centered around the Artificial Intelligence and Data Act (AIDA), Canada’s comprehensive AI legislation that was introduced in 2022 as part of Bill C-27. It failed to pass through Parliament before then–Prime Minister Justin Trudeau resigned in January 2025. The AIDA took a harm-based approach to AI and encountered criticism for its opaque development and vague provisions. Without an overarching AI law, the country has looked to existing laws (e.g., national and provincial privacy laws) to regulate AI and formed the Artificial Intelligence Safety Institute in late 2024 to study AI development and risks. The country also has a Voluntary Code that emphasizes the safe and responsible development and management of generative AI systems/tools. At the provincial level, Ontario’s 2024 Strengthening Cyber Security and Building Trust in the Public Sector Act sets guidelines for public sector use of AI with an emphasis on safeguarding personal information.

By the numbers

Of the 29 proposals we reviewed in North America, five specifically mentioned journalism; five addressed freedom of speech or expression; eight addressed manipulated or synthetic content; five addressed algorithmic discrimination and bias; five addressed intellectual property and copyright; 19 addressed transparency and accountability; eight addressed data protection and privacy; and four addressed public information and awareness.

Impacts on journalism and a vibrant digital information ecosystem

Public awareness and information provisions may improve journalists’ access to information, as well as their ability to share innovations — and what they mean — with the public. Much attention has been paid to transparency and accountability, with a particular focus on AI-generated content. News organizations have experience with disclosing uses of technology, and existing and proposed laws in the region that require disclosing AI-generated content are unlikely to pose a major burden. However, what kinds of disclosure are the most appropriate, effective and understandable remains an open question. 

Legislation in North America has also considered algorithmic discrimination and bias, which will shape how news organizations implement AI systems for hiring. Bias audits can allow for more in-depth examinations of AI tools and algorithms which yield opportunities for journalists to (1) better understand these systems and (2) explain them to the public. Similarly, news organizations will also need to comply with new data protection provisions. These personal data protections in AI systems, while important for personal privacy, may make it more difficult for journalists to investigate algorithms and alleged misuse of AI systems. 

These proposals also carry important implications for copyright and intellectual property. The North American region has still not fully agreed on how (and if) to handle compensation for digital content usage. Questions about training data and who owns the outputs of generative AI have wide-ranging implications for journalism and the digital information ecosystem. News organizations will be included in legal and legislative decisions surrounding copyright and intellectual property, which will shape compensation for digital usage and how news organizations develop and maintain their own AI systems. 

Finally, we observed several approaches to freedom of speech and expression in the region. Several U.S. states’ proposals acknowledged the First Amendment of the U.S. Constitution, which is pivotal to an open and free press. Others, relatedly, call out access to technology — and AI technology in particular, as in Montana’s Right to Compute Act — as a fundamental right, which could allow journalists greater opportunity to develop journalism-specific tools to aid in reporting and investigation.

Freedom of speech and expression

AI summary: North American AI proposals address freedom of speech and expression by suggesting protections for AI company employees, asserting existing First Amendment rights and viewing access to technology as a form of free expression. These approaches could strengthen journalistic reporting by supporting First Amendment rights and offering more opportunities for news organizations to develop AI models, though there are also potential challenges and unpredictable outcomes.

Approaches

Legislation in North America does consider freedom of speech, expression and information. The proposals we examined generally fit into three categories. 

1. Offering suggestions that AI companies provide protections to encourage their employees to voice concerns. New Jersey’s assembly resolution urges AI companies to provide “[c]urrent and former employees … the freedom to publicly report concerns until the creation of an adequate process for anonymously raising concerns.” This resolution is not legally binding and compliance is voluntary for AI companies. 

2. Acknowledging policies do not impede existing protections for individual freedoms. Certain approaches in the U.S. emphasize that First Amendment rights take precedence over any clauses of newly suggested (or signed) legislation (e.g., SB24-205 in Colorado and HB 149 in Texas). For instance, Colorado’s legislation states that nothing in the law “imposes any obligation on a developer, a deployer, or other person that adversely affects the rights or freedoms of a person,” and Texas’s legislation has a similar provision saying that nothing in the chapter on AI “impose a requirement on a person that adversely affects the rights or freedoms of any person …” In essence, these bills state that the legislation is not meant to impede First Amendment rights. 

3. Viewing access to technology as a form of free expression. One unique approach is in Montana’s SB 212 in which “[g]overnment actions that restrict the ability to privately own or make use of computational resources for lawful purposes, which infringes on citizens’ fundamental rights to property and free expression, must be limited to those demonstrably necessary and narrowly tailored to fulfill a compelling government interest.” Also known as the Right to Compute Act, Montana’s approach asserts access to technology and computational resources as fundamental rights to residents in the state.

Impacts on journalism

These policies impact journalism in several ways. For one, there are several examples of U.S. states supporting First Amendment rights, which should help to protect and bolster journalistic reporting, though these rights are not the primary focus of the legislation. New Jersey’s assembly resolution could also make it easier for journalists to report on allegedly nefarious and dangerous practices within AI companies. 

Montana’s “right to compute” approach suggests news organizations in the state may have more opportunities to develop AI models because using and building computational resources (including AI systems) are viewed as fundamental rights in this legal framework. This approach may allow for developments in AI that benefit journalism (e.g., developing novel content recommendation algorithms) by encouraging the idea that technology is a right for all. It is unlikely to be sufficient on its own: news organizations will also need extensive training and capacity building to achieve these goals. Moreover, the “right to compute” approach will likely make it more difficult to regulate the industry in other ways and may thus have wide-ranging repercussions that are difficult to predict.

Manipulated or synthetic content

AI summary: North American proposals on manipulated content aim to prevent harm by prohibiting nonconsensual intimate AI content, requiring consent for image/voice manipulation and mandating labels for deepfakes, sometimes with exceptions for satire or entertainment. And, in the United States and Canada, “false” information is protected constitutionally, subjecting policy proposals in this area to strict scrutiny. Indeed, most have been struck down prior to enactment or have quickly faced legal challenges. 

These types of proposals could protect journalists from deepfake misuse and offer opportunities for using synthetic media responsibly, but they could also limit reporting if definitions of “news organization” are too narrow or if consent rules are too strict and enable unintended censorship.

Approaches

There are many examples for how North American proposals have tried to address or are considering how to address manipulated content. These generally fit into four categories. 

1. Prohibiting the publication of nonconsensual intimate content created with AI. Several examples, such as the U.S.’s 2025 Take It Down Act, focus on mitigating harms from manipulated content by prohibiting the publication of nonconsensual “intimate visual depictions” of adults (and any “intimate visual depictions” of minors) that are either authentic or created with the assistance of computer programs, such as deepfakes (although many states already have laws prohibiting the publication of such content). The more original part of the act also requires platforms to establish a notice-and-removal process, and platforms must remove this content within 48 hours of receiving notice. 

2. Requiring consent for manipulating an individual’s image or voice. This approach protects an individual’s image and likeness from being used without authorization and provides property rights over one’s own personal image and likeness. For example, Tennessee enacted the Ensuring Likeness, Voice, and Image Security Act of 2024 (ELVIS Act), expanding the state’s statutory right of publicity in two ways.  First, it expands existing law to prohibit unauthorized commercial use of an individual’s voice, in addition to the existing restrictions on use of an individual’s name, photograph and likeness. Second, the ELVIS Act creates secondary liability for AI companies and platforms if they knew their tools or service facilitated the use of the person’s voice/name/photo/likeness and that the use was unauthorized.

3. Requiring labels or disclaimers for deepfakes. A number of proposals explicitly prohibit distributing deepfakes without disclosure (e.g., Hawaii’s SB 2687, New Hampshire’s HB 1432, South Dakota’s SB 164), especially regarding election-related content. Many of these types of proposals permit deepfakes if disclaimers/disclosures are readily presented with the content.

1. Providing carve-outs for satire, parody or journalism. A few proposals provide carve-outs for synthetic media that contain satire or parody (e.g., Illinois’s SB 150, South Dakota’s SB 164), as well as exceptions for news organizations that present synthetic media to audiences as part of official news coverage and disclose that it is non-authentic content. 

Impacts on journalism

Proposals that seek to address manipulated content in North America must strike a careful balance between countering disinformation and protecting free speech; “false” information is protected in Canada and the United States. Policy proposals in this area are subject to strict scrutiny, and if they do not maintain this careful balance, they could be misused for censorship, even in democracies. 

Legislation like the Take It Down Act, for example, will likely protect journalists’ identities from being recreated in nonconsensual intimate deepfakes; however, experts warn that the removal provision in the act poses serious risks to free speech and could be misused for political or ideological gain.

Similarly, several state-level deepfake laws have already been challenged or blocked in federal court. For example, The Babylon Bee, a satirical website, has filed a lawsuit challenging Hawaii’s SB 2687. 

There may be opportunities, as discussed in CNTI’s issue primers, in which deepfakes may serve as identity protection for both sources and journalists. Though there are carve-outs that allow for certain types of synthetic media — if clearly disclosed during official news reporting — further consideration should be given to how journalistic principles and synthetic media fit together. 

Labeling and disclosure requirements could be a relatively straightforward way to help the public better understand how AI-generated content is used in journalism, but such requirements could also be considered compelled speech under the United States’ First Amendment and could violate outlets’ editorial independence.

While some legislation includes provisions protecting satire and parody, as well as synthetic media covered by news organizations or used for legitimate reporting, the clauses determining who is eligible for those protections (i.e., what constitutes a news organization) may not cover all the ways journalism is conducted today. For example, Illinois’s SB 150 would protect AI-manipulated political communications content by a “bona fide newscast, news interview, news documentary, or on-the-spot coverage of a bona fide news event” if the AI content is clearly disclosed. However, it is unclear who determines which organizations count, especially since the exception language for this type of content explicitly states “a radio or television broadcasting station, including a cable or satellite television operator, programmer, or producer” is covered but does not necessarily include podcasts or internet sources (e.g., YouTube content creators). If content creators or independent journalists are not included, these bills could be used to limit their reporting.

Laws that require consent, such as Tennessee’s ELVIS Act, may require news organizations to more closely analyze the advertisements they run to avoid being held liable for any voice or image manipulations. This is because the final ELVIS Act increases liability for media companies by including a narrower fair use provision in the final act than what existed in preceding drafts. 

As many proposals require disclosure of synthetic media content, news organizations can serve as valuable collaborators along with other stakeholders to design and implement appropriate and effective labels.

Algorithmic discrimination and bias

AI summary: North American proposals address algorithmic discrimination and bias by focusing on consumer and employment protections, safeguarding vulnerable groups and requiring regular bias audits for AI systems. These approaches may affect journalism by requiring news organizations using AI for hiring or content recommendations to conduct bias audits and ensure fairness, while at the same time potentially providing journalists with valuable data for reporting on AI.

Approaches

Several proposals we reviewed in North America consider algorithmic discrimination and bias. These generally fit into the three categories. 

1. Emphasizing consumer and employment protections. For example, Colorado’s SB24–205 strives to protect people from algorithmic discrimination by placing regulations on “high-risk” AI systems — and their developers and deployers — that make “consequential decisions” such as accessing education, employment, healthcare or insurance resources. Both developers and deployers of these high-risk AI systems in Colorado must practice reasonable care to mitigate algorithmic discrimination, and the legislation also requires risk assessments and annual reviews.¹ 

2. Protecting vulnerable populations and groups from algorithmic discrimination and bias. These types of provisions are found in Guam’s Bill No. 64-38, Hawaii’s SB 59, Texas’s HB 149 and Canada’s AIDA (failed). Guam’s proposal seeks to create the “Guam Artificial Intelligence Regulatory Taskforce,” which would be responsible for developing a regulatory framework to prevent algorithmic bias. 

3. Requiring bias audits related to algorithm usage. New Jersey’s A3855, for example, requires bias audits of “automated employment decision tools” to mitigate the potential negative impacts of algorithmic discrimination. A bias audit in New Jersey would use demographic variables (e.g., income, age, gender, race, ethnicity, religion) to measure scoring and selection rates of the employment tool based on the training data of employers or employment agencies that use the tool. These types of AI employment tools generally require regular bias audits (often yearly).

Impacts on journalism

These proposals may impact journalism in a variety of ways. To start, organizations that use AI systems during the hiring process would likely be impacted. News organizations in New Jersey, for example, would likely be required to produce bias audits demonstrating that their usage of AI complies with regulations and does not adversely harm protected groups. 

It is unlikely that these audit proposals would require audits to be made public. In New Jersey, for example, it remains to be seen where and how independent auditors would release the results of their bias audits. That said, on the off chance that the audits are made available to journalists, they could serve as a valuable resource for reporting on AI tools, companies and technological developments. 

Regulation on algorithmic discrimination and bias may also affect how news organizations can digitally interact with customers. For example, while New Jersey’s law only monitors automated employment decisions, broader legal approaches might require news organizations to verify that article recommendation algorithms do not disadvantage readers or provide different information based on protected attributes (e.g., race, ethnicity, gender). Advertising decisions that include AI models would also fall into this category.

Intellectual property and copyright

AI summary: Several proposals in North America are considering intellectual property and copyright, with some states clarifying ownership of AI-generated content, upholding existing laws and requiring disclosure of training data. These approaches will influence how news organizations develop their own AI models and how they handle content for training purposes, as ongoing legal battles will determine the future of intellectual property and copyright in the region.

Approaches

Intellectual property and copyright regarding AI developers’ training data has received a lot of attention in the U.S. judicial system. For example, in June 2025, Anthropic won an initial judgement that the company’s  scanning of books to train generative AI models constitutes fair use. This ruling sets a precedent that would likely impact the anticipated New York Times and OpenAI case. 

On the policy front, there are just a few examples of individual U.S. states considering the relationships between intellectual property and copyright and AI; copyright is usually determined at the federal level in the U.S. These state-level efforts generally fit into three categories. 

1. Specifying the ownership of AI-generated content. For example, HB 1876 in Arkansas has attempted to clarify who owns material that has been created with generative AI models: “the person who provides the input or directive [owns the generated content] … provided that the content does not infringe on existing copyrights or intellectual property rights.” 

2. Providing language that upholds existing laws on the topic. For example, Montana’s Right to Compute Act, while innovation-focused, has a provision in Section 5 to uphold “federal and state intellectual property laws” but does not clarify who owns content created by generative AI models. 

3. Requiring the disclosure of training data and the ownership of that data. These disclosure requirements aim to protect intellectual property. Some of the requirements of AI developers in California’s AB 2013 include public transparency around:

• The sources or owners of the datasets.
• Whether the datasets include any data protected by copyright, trademark, or patent, or whether the datasets are entirely in the public domain. [Note: California’s legislation does not include any penalty for using these types of data.]
• Whether the datasets were purchased or licensed by the developer.

Impacts on journalism

In the U.S., state-level copyright is preempted by federal copyright provisions, so any state-level activity would ultimately be subject to what happens at the federal level. On that front, we are likely to see more consequential activity as ongoing lawsuits and court cases conclude. That being said, the lack of activity could simply be noted as deference to federal law. Separately, a series of papers from the U.S. Copyright Office released in 2024 and 2025 provided recommendations for how U.S. copyright law could be updated to account for generative AI; however, because this report focuses on policies, strategies and legislation, these papers were not included in our analysis. 

It remains to be seen whether training models with news organizations’ content is determined to violate intellectual property, copyright and fair use laws, though rulings to date suggest it does not. These types of regulations will likely shape how news organizations can legally develop their own AI models for content recommendation, personalization and content creation. This will largely be based on the types of training data that are available outside the ownership or purview of the news organization, depending on the outcomes of existing lawsuits and court cases. The ongoing legal battles between news organizations and technology companies over AI developers using news content for training purposes will shed light on the central topics of intellectual property and copyright, especially about who owns data and what data can be used to train AI models.

Amid these uncertainties, several news organizations have also signed deals with AI developers to grant access to their news archives, while other news organizations are hesitant to do so. Determining who owns AI-generated content across the region, especially as so few legislative examples exist, is critical. And overall, the region has still not fully grappled with how to value digital content, which CNTI wrote about in detail in our analysis of media remuneration policy. 

Transparency and accountability

AI summary: North American AI proposals on transparency and accountability focus on requiring disclosure of AI-generated content, especially in elections, as well as mandating transparency around training data and AI models, algorithmic impact assessments and the development of AI detection tools. These approaches will require news organizations to label AI-generated content, potentially develop AI detection tools and will enable journalists to report on AI systems in more detail due to increased transparency.

Approaches

Transparency and accountability have received a lot of attention in North America during the last several years. The range of transparency considerations is extensive. While nearly all of the proposals we examined are legally-binding legislation (or would be if passed), New Jersey passed a non-binding assembly resolution urging whistleblower protections for employees at AI companies. 

The proposals we reviewed generally fit into five categories. 

1. Requiring disclosure of AI-generated content. Several of these approaches revolve around elections and election communications (e.g., advertising). Several approaches also define acceptable types of disclosure (e.g., SB2687 in Hawaii, SB150 inIllinois, HB1432 in New Hampshire, SB164 in South Dakota), which often require explicit labels or disclaimers denoting AI-generated content. A slightly broader but related approach occurs in Ontario’s Bill 194, which stresses that public service entities need to disclose their use of AI systems (with specific provisions made by the Lieutenant Governor in Council).

2. Requiring training data and model transparency. For example, California’s AB 2013 and Washington’s HB 1168 outline the required documentation that AI developers need to share about training data for AI systems, including (1) descriptions of the data, (2) whether personal data were included, (3) the time period the data were collected, etc. Relatedly, SB 59 in Hawaii includes a provision requiring anyone who uses AI for an algorithmic eligibility determination — in other words, determines eligibility for an important opportunity (e.g., employment, insurance) using an algorithm — must issue a notice about how personal data are used in those decisions. 

3. Requiring algorithmic impact assessments, especially for “high-risk systems.” SB24-205 in Colorado states that these assessments should include “a description of any transparency measures taken concerning the high-risk artificial intelligence system, including any measures taken to disclose to a consumer that the high-risk artificial intelligence system is in use …” as a way to satisfy the conditions of the law.

4. Requiring disclosure of non-human AI interactions to consumers. HB 516 in Alabama, HP 1154 in Maine and SB 149 in Utah aim to (1) prevent deception in online trade and commerce and (2) strengthen consumer protections. For example, users would need to be notified they are engaging with an AI agent and not a human. 

5. Requiring development of AI detection tools. California’s AI Transparency Act (SB 942) requires AI providers to release an AI detection tool for users. The law will “require a covered provider, as defined, to make available an artificial intelligence (AI) detection tool at no cost to the user that meets certain criteria, including that the AI detection tool is publicly accessible.” The tool will need to provide users with information about whether content they provide to the tool was altered by the provider’s AI system, thus aiming to increase transparency and accountability. 

Impacts on journalism

These types of proposals impact journalism in a number of ways. Disclosure of AI-generated content will require news organizations to incorporate labels prior to publishing. Identifying the most effective types of labels for specific types of content is still an open question and not explicitly outlined in the existing legislation. Based on California’s SB 942, news organizations that provide generative AI tools for public use (e.g., chatbots trained on their own content) would be required to develop a detection tool if the organization reaches the threshold of 1 million monthly users.

Mandatory labeling also carries implications for free speech in the sense that requiring labels could be considered compelled speech. In other words, these requirements could be interpreted as the government telling individuals (i.e., news organizations and journalists) what they must say, which would be a violation of the U.S.’s First Amendment. 

Proposals that include transparency and accountability decisions will also allow journalists to report on AI systems in more detail. For example, increased transparency about training data allows for greater evaluation of AI models and their outputs.

The two proposals we examined that directly focus on news organizations (A5164 in New Jersey and S6748 in New York) carry implications for common journalistic tasks. One of these is journalists’ use of AI transcription and translation, which can be considered generative AI, depending on the specific tool used. In New Jersey’s bill, for example, these uses would need to be disclosed to audiences. Journalists and newsrooms will likely need to further consider how much (and what) information needs to be provided to audiences if, for example, generative AI was used early in the reporting process to transcribe and summarize meetings. 

Lastly, the impacts will be shaped by whether the proposals fall more heavily on developers or deployers. For example, California’s AB 2013 and Washington’s HB 1168 place transparency requirements on AI developers, which should enable greater journalistic coverage of AI systems. On the other hand, legislation that focuses on deployers would apply to news organizations and journalists that use AI systems, holding them responsible for the outputs of these systems. There are examples, though, of approaches that include a focus on both, like Colorado’s SB24–205, which outlines transparency requirements when using high-risk AI systems.

Data protection and privacy

AI summary: North American AI proposals on data protection and privacy focus on requiring personal data to be de-identified and ensuring secure handling of anonymized data. These approaches will require news organizations to comply with new regulations if they use personal data in their AI systems. They might also make it harder for journalists to evaluate AI systems if training data are protected.

Approaches

Several proposals consider the implications of data protection and privacy when building and deploying AI. The proposals we reviewed generally fit into the following categories. 

1. Requiring personal data to be deidentified. There are several provisions for protecting “personal provenance data,” such as SB 942 in California, and personal information, such as SB 59 in Hawaii. These include any metrics or information that can be traced back to an individual user. Relatedly, Utah’s 2024 SB 149 (whose expiration date was extended in 2025 by SB 332), includes “deidentified data” to prevent personal data from being used to identify individuals. 

2. Requiring secure handling and processing of anonymized data. Canada’s AIDA outlined requirements for those processing and/or working with anonymized data which include (1) how the data are anonymized, (2) how the data are used and (3) how the data are managed. 

Impacts on journalism

These approaches impact journalism in several ways. News organizations may need to comply with regulations if they are developing AI systems using personal data, although most current laws focus primarily on regulated industries. There will also be limits on how data (and what types of data) can be shared between news organizations should they collaborate to develop and deploy AI systems for their specific needs.

Requirements for data protections are likely useful for protecting individual privacy, but journalists may encounter difficulty when examining and investigating AI systems if the training and testing data are protected due to the use of personal data. Learning what types of deidentified data are acceptable for accessing AI systems can yield an opportunity for journalists to evaluate training data without handling personal information.

Public information and awareness 

AI summary: North American AI legislation regarding public awareness mainly involves creating public campaigns, training government employees and funding educational programs, but these efforts rarely include journalists. Journalists could play a key role in raising public awareness about AI and would benefit from learning about AI systems to better report on technological developments.

Approaches

There are few examples of legislation in North America that explicitly focus on public information and public awareness, but S. 1699 in the U.S. Congress is noteworthy. The proposals we reviewed generally fit the following categories.

1. Developing public awareness campaigns. The U.S. Senate has introduced S. 1699, which would create a “public awareness, education, and consumer literacy campaign” to provide information about AI technologies. The Secretary of Commerce would be responsible for overseeing the campaign, which would also include a requirement to measure public literacy on AI technologies.

2. Creating training programs for certain government employees. Texas’s Responsible AI Governance Act places an emphasis on “training programs for state agencies and local governments on the use of artificial intelligence systems.” The scope of training programs does not extend to media organizations more broadly, or beyond government.

3. Providing funding for AI courses and training. This includes educator training resources for educational institutions. These efforts are found in several U.S. states and range from creating commissions to study the impacts of AI in the classroom, such as in HCR 66 in Louisiana, to funding the development of a state pilot program that will create an AI tool for classroom instruction and train the educators who will use said tool, like in Sections 143 and 144 of Connecticut’s HB 5524. However, the scope of these educational programs does not extend to journalists or news organizations. 

Impacts on journalism

Journalists may have a role to play in raising public awareness — particularly as potential leaders in a national AI awareness and literacy campaign as outlined in the U.S. bill S. 1699. Journalists will thus need access to information about AI systems and models to build audience knowledge and understanding of these technological developments.

While several states’ proposals stress the importance of public awareness of AI in educational settings, the current examples of education provisions that aim to increase awareness do not include journalists specifically. Yet, there may be opportunities for news organizations to partner with schools to promote education initiatives for school-aged individuals and the general public more broadly. 

Outside of S. 1699 in the U.S. Congress, which proposes a national AI awareness campaign and highlights the AI topics the public can learn about (e.g., machine translation, content provenance), the legislation we examined did not provide a clear description of what information is important for the public to know. Still, public awareness campaigns help journalists learn about AI systems. This can assist them in their professional role of delivering information to the public as well as in reporting on developments in AI and automation technologies.

Conclusion

AI legislation has received a tremendous amount of attention in the United States over the last two years, driven in large part by state activity. Less legislation is being considered in Canada; however, it is difficult to compare the two countries given disparities in population and the number of U.S. states versus Canadian provinces and territories.  Discussions in both countries are beginning to include important concepts related to journalism, namely transparency, accountability, data protection and free expression; however, they rarely name journalism explicitly, with a few noted exceptions  in New Jersey and New York. Whether named directly or not, it is important that policymakers consider the ways legislation could harm an independent news media and an open internet. 

Within the journalism community itself, there are opportunities to set standards on transparency around use of AI-generated content. There are also several examples of provisions that protect journalistic coverage of deepfakes. Yet, there are also shortcomings in the legislation outlined above. All in all, the existing patchwork of approaches risks inconsistency in data protection, disclosure requirements and protections for freedom of speech/expression. 

Federal and state legislation in the United States is an area to watch in the near future. It is also worth keeping an eye on future developments in Canada, especially given the country recently formed a new government that might have a different take on AI legislation. Early signs point to Canada presenting an updated AI regulatory framework in the near future that takes copyright into consideration.

1.  Colorado introduced SB25–318 in Spring 2025 which would revise SB24–205 by only covering larger AI developers and solely focusing on existing anti-discrimination laws. ︎

The post North America appeared first on Center for News, Technology & Innovation.

Representatives from 15 Latin American and Caribbean countries agreed to adopt a regional approach and roadmap to AI at a UNESCO summit in late 2024. These documents take a human-centered approach that foregrounds the importance of both multi-stakeholder governance and national sovereignty. Chile and Brazil, which lead the region in measures of AI readiness, are positioning themselves to set policy standards in their respective languages. Both have well-developed national strategies and are actively creating comprehensive legislation. Yet Brazil has long acted somewhat independently of its regional neighbors, and it remains to be seen whether other countries will follow its lead.

While the major legislative proposals in these two countries take a balanced approach — citing the EU AI Act and borrowing liberally from its risk framework, while simultaneously creating sandboxes and incentives for development — technology companies have criticized both Chile and Brazil’s proposals as excessively burdensome. Meanwhile, the national laws that have successfully been enacted to date (in El Salvador and Peru) include components focusing on both innovation and harm, primarily encourage innovation and investment, and contain relatively little detail about consequences for violations. Moreover, few countries have capacity to develop appropriate oversight mechanisms, so it is unclear how consequences would be enforced.

In Brazil, one state “leapfrogged” the national process to pass the country’s first AI law, which takes a more innovation-focused approach than the proposed federal legislation. The federal bill, first proposed in 2023, passed the Senate in January 2025 and remains under discussion in the House of Deputies. (Specific harm-based laws addressing data centers, deepfakes and online safety were also proposed in 2024, although these are at a much earlier stage in the legislative process.) A similar comprehensive law, proposed in 2024, is currently being discussed in Chile.

Five policies we reviewed in this region mention journalism or journalists, more than any other region we reviewed. Most of those mentions are brief, either stating that new AI technologies are worsening disinformation or providing carve-outs for the sector from particular policy clauses, such as consent requirements for synthetic content or strict licensing requirements for the use of copyrighted materials.

By the numbers

Of the 80 proposals we reviewed in Latin America and the Caribbean, five specifically mentioned journalism; seven addressed freedom of speech or expression; 24 addressed manipulated or synthetic content; 39 addressed algorithmic discrimination and bias; 18 addressed intellectual property and copyright; 48 addressed transparency and accountability; 43 addressed data protection and privacy; and 38 addressed public information and awareness.

Impacts on journalism and a vibrant digital information ecosystem

Given how few laws in this discussion have been enacted to date, the impacts of AI regulation on journalism in Latin America and the Caribbean are still an open question. However, the information space appears to be more front of mind in these discussions compared to those in other regions. It is important to ensure that legislative language accomplishes its intended effect and does not raise new risks to the independence and diversity of a vibrant information ecosystem. 

Ecuador’s proposed Organic Law for the Regulation and Promotion of AI in Ecuador (2024), one of the most comprehensive in the region, stands out for its broader vision of a vibrant and diverse digital information environment. Article 31, which focuses on “diversity and plurality in digital environments,” requires AI content recommendation systems to “expose users to a diversity of sources, topics and perspectives” and “facilitate equitable access to public-interest content from local, community and independent media.” Article 32, on “ending algorithmic censorship and manipulation,” requires additional transparency and appeals processes in this context. Still, it is not yet clear how effective these provisions will be at meeting their stated goals.

Freedom of speech and expression

AI summary: Freedom of speech is rarely included in legally binding articles of Latin American and Caribbean AI policies, with approaches focusing on banning AI systems that don’t respect these freedoms or trying to balance free speech with fighting false information. It’s unclear how these rules will affect the information environment, though they could potentially support media independence.

Approaches

Freedom of speech and expression are sometimes mentioned in the preamble or justifications of legal proposals, but they are almost never addressed in the legally binding articles of these proposals: only seven out of 80 documents we reviewed in the region addressed this topic. Moreover, about half of them simply affirm its importance without addressing it in enforceable ways.

Proposed laws that emphasize freedom of speech or expression typically address it in one of two ways.

1. Banning or placing additional restrictions on AI systems that do not respect these freedoms. At least two proposals take this approach. Article 19 of Argentina’s 2024 S-Bill 2573-S-2024 bans the use of AI systems that “violate fundamental human rights” such as privacy, freedom of expression, equality or human dignity.” Similarly, Colombia’s Bill 442/2025 considers any “AI system that may affect the exercise of rights to personal privacy, freedom of expression, transparency or access to public information” to be a high-risk system. In this framework, high-risk systems are subject to higher levels of scrutiny and restriction.
   
2. Balancing freedom of speech with the need to control disinformation. Article 32 of Ecuador’s 2024 “Organic Law for the Regulation and Promotion of AI in Ecuador” places some limits on content recommendation algorithms in order to “minimize restrictions on freedom of expression.” These limits include clear terms, human supervision, accountability structures and appeal processes, as well as sector-wide best practices. A similar clause in Article 4 of Costa Rica’s 2023 proposed “Law for the Regulation of AI in Costa Rica” (Bill 23771) emphasizes that Costa Rica’s constitution consecrates freedom of expression, “which includes the freedom to look for, receive, and share information.” Uruguay’s 2024-2030 national AI strategy includes a similar juxtaposition.

Impacts on journalism

It is hard to predict how these proposals would impact the information environment. Restrictions on content recommendation algorithms could potentially improve the information environment created by current social media content algorithms, but they also raise questions about how the government will define “digital platforms” and “communications media.” Article 32 of Ecuador’s “Organic Law for the Regulation and Promotion of AI in Ecuador” (2024) emphasizes sector-wide standards and best practices, which could support media independence — but it will be difficult to assess its impacts without more knowledge about who will participate.

Manipulated or synthetic content

AI summary: Many Latin American and Caribbean countries are creating laws about manipulated content, mainly by requiring labels, watermarks or consent, but there is variation in who is responsible and when exceptions are allowed for journalism. These labeling and consent rules could help journalists deal with online harassment and improve information quality, but outright bans on deepfakes might create problems for journalistic work.

Approaches

Most Latin American and Caribbean countries with laws, bills or strategies have at least some mention of manipulated or synthetic content: Argentina, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Mexico, Panama, Paraguay, Perú, Uruguay and Venezuela all do — while Bolivia, Ecuador, El Salvador and Jamaica do not. Out of the 80 documents we reviewed, 24 addressed this issue. 

Latin American and Caribbean laws, policies and proposals generally offer four approaches to synthetic and manipulated content — both as standalone proposals and as packages.

1. Labeling manipulated content. Such laws differ in whether they put responsibility on software developers, users creating the content or anyone sharing manipulated content. For example, Chile’s 2023 Bill 15869-19, proposed by the Congress and later combined with Article 10 of an executive proposal, aims to regulate AI, robotics and associated technology and focuses on developers and users, requiring “developers, providers and users of AI systems that generate or manipulate image, sound or video content that noticeably resembles existing people, objects, places, entities or events” to ensure that anyone who sees the content is aware that it has been generated by AI. Article 14 of Paraguay’s 2025 “Bill that promotes the use of AI for the social and economic development of the country” says it will develop penalties (which could include fines, suspension of operations, or civil or criminal liability) for sharing manipulated content without consent, which could potentially impact people who did not themselves know that the content was manipulated.
2. Requiring explicit consent from people or entities whose image is being manipulated. Brazilian Bill 5721/2023 proposed in 2023 that primarily focuses on deepfake nudity and pornography specifies in Article 3 that “the creator of inauthentic synthetic content must obtain the previous consent of persons whose images or voices will be used,” in addition to labeling. Some proposals place the responsibility on developers by forbidding the commercialization, sale, distribution and/or use of systems that have the functionality to manipulate images and audio if those systems do not include mechanisms for verifying consent. Mexico’s December 2024 Bill for National Law that Regulates the Use of Artificial Intelligence, for example, says in Article 40 that developers must revise models so that they cannot generate illegal content and notes in Article 43 that AI-generated content and images must be labeled visibly. Several countries have also passed broader laws against digital violence that include but are not limited to synthetic images, such as Mexico’s Ley Olimpia, which entered into force in 2021. (A detailed analysis of broader laws against digital violence is beyond the scope of this report.) 
3. Updating existing penal codes. Several laws, such as Peru’s Law 32314 (passed in April 2025), also update penal codes to clarify that existing laws about child pornography, libel or defamation also refer to AI-generated content.
4. Making exceptions for journalistic uses. Some laws provide clear exceptions to consent provisions for journalism, satire, educational purposes or branding. Very few labeling provisions include exceptions. Article 36 of Argentina’s Bill 2130-D-2025, which requires consent via an amendment to the Civil Code, makes an exception for “a priority scientific, cultural, or educational reason” with precautions to avoid unnecessary harm which might include reporting on synthetic content.
5. Banning the creation of deepfakes entirely. Dominican Republic Bills 495 and 563, both proposed in 2025, would make using AI to generate and spread disinformation punishable by both prison terms and fines.

Impacts on journalism

Labeling and consent requirements on synthesized audio, images and video have strong potential to support journalists in responding to online harassment. Both types of requirements — either alone or in combination — make this form of harassment more clearly illegal. These requirements also create avenues for journalists to dispute harassing content and have it removed from online portals and platforms. There remains debate over who should be legally accountable for violations: if developers are accountable, liability will rest with a smaller number of parties who are easier to identify; on the other hand, end users have more proximate responsibility for their creations.

By and large, labeling requirements on synthesized audio, images and video have potential to improve the overall information ecosystem. Legitimate journalistic uses of the technology, such as anonymizing faces in a video or reporting on deepfakes, are unlikely to be harmed by these requirements — they can simply be labeled, and often already are. Consent provisions are also unlikely to impact uses that support anonymity, since sources would presumably be informed. However, some consent provisions could make reporting more difficult, particularly if they prohibit journalists from showing manipulated content that is the subject of reporting. Explicit carve-outs may be effective. Banning deepfakes altogether may potentially have negative consequences for journalism. Legitimate uses — such as using avatars to protect journalists and sources, or including a brief clip to debunk a political deepfake — could also fall under the bans and potentially make journalists and their sources less safe.

Algorithmic discrimination and bias 

AI summary: Latin American and Caribbean countries are addressing algorithmic bias with a human rights-centered approach — such as prohibiting high-risk systems, mandating audits and diverse training data, and strengthening end-user rights. This approach presents opportunities to hold AI developers and deployers accountable.

Approaches

Every single country in the region with a law, bill or strategy includes at least one reference to algorithmic discrimination and bias. Across countries, non-discrimination and bias prevention show up as broad principles for AI laws and frameworks. Every country has at least one document that mentions this topic; 39 out of 80 documents address this topic.

Not all 39 documents that make reference to algorithmic discrimination and bias address it in a meaningful way. Those that do use a number of different approaches.

1. Banning AI for uses where the likelihood of bias and its impacts are both high. Proposals in Argentina in 2024, Bolivia in 2025, Brazil in 2023, Chile in 2023, Costa Rica in 2024, Ecuador in 2024 and Paraguay in 2025 include a risk classification system most likely borrowed or adapted from the EU AI Act, which prohibits most forms of biometric surveillance and social credit systems, among other uses.
2. Ensuring representation earlier in the development process. These include proposals to encourage more diversity among working teams. Chile’s 2025 Decree 12 requires more representative training data. Venezuela’s 2025 AI Bill and the 2024-2028 Brazilian AI Plan incentivize better language models. It is unclear how these proposals would be enforced, although they sometimes exist in combination with auditing requirements.
3. Mandating auditing and public registries. A number of proposals would make participation in periodic bias audits a requirement for doing business in their country, with penalties for unsatisfactory results ranging from warnings to large fines to potential loss of business licenses. For example, Article 17 in Paraguay’s 2025 proposed “Bill that promotes the use of AI for the social and economic development of the country” would create a national registry of AI systems, prohibit the use of biased or insufficient data, require traceability and explainability mechanisms and implement regular auditing. Organizations that fail their audit will be required to create a remediation plan and may be subject to additional fines and punishments. Some proposals, such as Article 29 in Ecuador’s 2024 “Organic Law for the Regulation and Promotion of AI in Ecuador,” further highlight the need to mitigate bias against specified historically marginalized groups.
4. Updating existing laws about personal data and discrimination. Several countries are updating existing laws to clarify who is legally responsible for discriminatory AI and requiring AI tools to comply with these laws. Most of these laws hold both the developer and the deployer accountable. In this vein, Argentina’s 2025 proposed 2130-D-2025 would add considerable text to the country’s Personal Data Protection Act by specifying who is responsible for violations of data policy and outlining rights for anyone impacted by discriminatory automated decisions.
5. Strengthening end-user rights. Another relatively common proposal enshrines end-user rights into law, including the right to human review and human decision-making. That is, many laws guarantee that people can appeal algorithmic decisions and request that a human review the decision. Article 51 of the 2025 Bill 495 from the Dominican Republic, for example, grants the right to appeal to anyone affected by decisions made partially or fully by an AI system, as long as one of the following conditions is met: the decision violates fundamental rights, includes discriminatory bias, creates harm in any area regulated by the state, lacks sufficient transparency or greatly harms the user. If this type of law is not accompanied by larger-scale auditing, problems would primarily come to light when individuals report impacts and seek human review. That could mean that the responsibility for correcting discrimination would lie primarily with the people who are impacted by it.
6. Ensuring diverse outputs of recommendation algorithms. In addition to clauses about algorithmic bias more broadly, Ecuador’s 2024 “Organic Law for the Regulation and Promotion of AI in Ecuador” also includes two articles addressing “diversity and plurality in digital environments” and “algorithmic censorship and manipulation.” Article 31 requires providers of content recommendation algorithms to ensure that users see diverse sources and perspectives, including “public-interest content from local, community and independent media.” Article 32 specifies that these platforms must have human supervision, traceability mechanisms and an appeals process if content is removed or has limited visibility.

Impacts on journalism

Most of the legal proposals addressing algorithmic bias in Latin America and the Caribbean are primarily concerned with decision-making algorithms and are unlikely to apply directly to journalistic uses of AI. 

Articles 31 and 32 of Ecuador’s 2024 “Organic Law for the Regulation and Promotion of AI in Ecuador,” which focus on content recommendation, are the primary exception. Article 31 requires algorithms to provide diverse sources. Doing so could potentially improve the information environment created by current social media content algorithms, as well as raise the visibility of journalistic content online. However, until it is operationalized, there remain questions of how it could negatively impact journalism. Article 31, as currently written, may require producers of journalism to include “a diversity of sources” — not just the platform and search companies. If so, producers of journalism may be unable to use recommendation or prioritization algorithms on their own apps or platforms because they are restricted from offering content exclusively from their own outlet or organization. Moreover, it is unclear who will assess the “diversity of sources, topics and perspectives” or determine what counts as “public-interest content.” Without clarity, these terms could potentially be weaponized against content producers. Article 32, which restricts the use of content recommendation algorithms and requires human supervision and appeal mechanisms, explicitly applies to “communications media.” Even once these definitional questions are clarified, Ecuador may not have the leverage to enforce these laws on multinational social media platforms. Most likely, these companies would either pull out of Ecuador or refuse to comply, as they have done in the face of regulation elsewhere. 

Provisions in other policies might also apply to recommendation, personalization or content-delivery algorithms. In these cases, algorithms may require additional scrutiny or transparency from news organizations, but they do not seem likely to foreclose legitimate journalistic uses.

In addition, the role of registries and audits may present an opportunity for journalism to report on the accountability of AI developers and implementers. In general, so-called “sunshine laws” that promote transparency in particular sectors have helped journalists access information.

Intellectual property and copyright

AI summary: Countries in Latin America and the Caribbean are taking varied approaches to intellectual property for AI, focusing on the rights of copyright holders if their work is used to train AI systems, the intellectual property of developers and/or ownership of AI-generated works. These rules could help news organizations financially by making AI developers pay for copyrighted content, but it’s important to have exceptions for journalistic uses of AI, especially since laws that punish end-users could put journalists at risk.

Approaches

Countries in Latin America and the Caribbean rely on intellectual property law to protect Indigenous and traditional arts from appropriation and theft. This salience may be why intellectual property and copyright come up across most countries with AI laws, strategies, or proposals in the region; the exceptions without such provisions are Bolivia, Paraguay and Venezuela. Because each country has a somewhat different intellectual property regime, it is challenging to summarize comprehensively. For example, Colombia has collective management of copyright, with several large organizations empowered to administer rights on behalf of creators. Out of the 80 documents we reviewed, 18 address this concern.

Proposals in Latin America and the Caribbean include six approaches.

1. Participating in international regulation and standards. Chile’s Decree 12, updating their national AI policy as of January 2025, recommends “participating in international dialogues and decision-making about regulating AI in relation with intellectual property, contributing to the formation of global policies.” 
2. Requiring prior authorization to use copyrighted work to train AI models. A number of proposals emphasize protecting existing copyrights, often through a combination of authorization, licensing and transparency mechanisms. For example, Article 6 of Costa Rica’s proposed “Law for the Implementation of AI Systems” (Bill 24484/2024, one of three bills under discussion) includes “the use of content protected by copyright and associated rights, which in all cases will be subject to the previous authorization of the corresponding rights-holders” as an area of primary impact of new AI technology, and thus subject to state evaluation and authorization. Article 13 of the same law explicitly prohibits unauthorized use of copyrighted materials, and Article 14 imposes a transparency requirement about the use of copyrighted materials. Ecuador’s 2024 “Organic Law for the Regulation and Promotion of AI in Ecuador” protects materials in the public domain from claims of copyright if an AI re-generates them. Colombia, Mexico and Panama have similar provisions in bills currently under discussion.
   1. Including exceptions for fields like research and journalism. Some laws contain provisions that exempt the use of copyrighted materials for scientific or journalism purposes. For example, the comprehensive law currently under discussion in the Brazilian Chamber of Deputies, Article 42 of Bill 2338/2023, includes this provision: “The automated use of works, such as extraction, reproduction, storage and transformation in processes of data or text mining in AI systems, does not constitute a copyright offense in activities undertaken by organizations and institutions of research, of journalism, or by museums, archives, or libraries, as long as … 1. Its objective is not the simple reproduction, exhibition, or dissemination of the original work; 2. The use is in the necessary amount for the goal to be reached; 3. It does not unjustly prejudice the economic interest of the title-holders, and; 4. It does not interfere with the normal use of the works.” This type of provision would make it allowable to use AI to recognize patterns in published books (without author consent), but not to train commercial generative AI on those same books.
3. Requiring traceability of sources or disclosure. Often in combination with prior authorization and licensing requirements, some proposals require text generation to be clearly traceable by including cited sources. Panama’s 2024 proposed “law which establishes a legal framework, promotion and development of AI in the Republic of Panama,” which is being discussed, lays out the need for traceability in a discussion of guiding principles in Article 13 and is explicit that the use of AI does not exempt anyone from responsibility if they have violated someone else’s intellectual property in Article 9.
4. Explicitly protecting AI models, datasets and algorithms as intellectual property. To support innovation, a number of provisions focus on the intellectual property of model developers and creators. A proposed Colombian law presented to the Congress in May 2025, includes a provision that the Ministry of Science, Technology and Innovation will provide legal and technical assistance so that AI researchers and developers can take advantage of patent and copyright protection. The 2025 Argentinian Bill 0511-S-2025 would require government agencies to prioritize the intellectual property of the companies it regulates. Proposed Dominican and Salvadoran laws also emphasize the intellectual property rights of developers.
5. Clarifying the copyright status of AI-generated works. Who owns work that is partially or wholly generated by AI is not uniformly agreed upon throughout the region. One provision offers potential solutions: Ecuador’s 2024 “Organic Law for the Regulation and Promotion of AI in Ecuador” says that work generated with help from AI can be copyrighted by its human author in Article 34, while Article 35 says work generated fully autonomously by AI goes directly to the public domain. Mexico also has a 2025 proposal addressing the copyright status of AI-generated work in which the law details a category of “significant human interaction” that creates works protected by copyright, while works created without it belong to the public domain.
6. Updating existing laws. A few proposals attempt to address these issues by updating existing copyright laws, but many of them (such as Peru’s Law 32314, passed in April 2025) lack specificity.

Impacts on journalism

The implications for journalism are particularly difficult to tease out given the many complexities of extant copyright law and its overall inadequacy for the digital information ecosystem. None of these proposals are specific about how to assess the value of digital content. That said, international cooperation and regulation could go a long way towards simplifying the current patchwork of intellectual property regulation, especially if there is diverse representation from this region. 

On the whole, requiring developers to license and pay for copyrighted content they use to train their models is likely to benefit news organizations financially. Exemptions for journalistic uses of AI tools are also an important safeguard, especially in countries where laws penalize end-users, which means that journalism organizations may be at risk of inadvertent violations if they use third-party tools. That said, carve-outs will then need to specify who and what qualifies for them.

Transparency and accountability

AI summary: AI policies in the Latin American and Caribbean region are guided by the principles of transparency and accountability, but not all proposals operationalize them; those that do include mandatory disclosure of AI interactions, a right to human review, requirements for accessible language in documentation and the creation of public registries. In some cases, it remains unclear whether the burden of compliance will fall on developers or the organizations that deploy the technology, and some rules might place a heavy burden on news organizations.

Approaches

Algorithmic transparency and accountability come up in laws and legislative proposals across Latin America and the Caribbean as overarching principles of AI governance and regulation. In fact, every country with at least one bill, law or strategy referenced this policy component. It was also the most commonly addressed topic in the region, appearing in 48 of the 80 documents we reviewed.

In this region, there are six main approaches to address transparency and accountability.

1. Requiring that users must know that they are interacting with an AI system. Such provisions range from requiring watermarking or labeling on AI-generated content to requiring labeling on all algorithmic systems and tools. The latter would require apps and platforms (including social media, search engines and streaming sites) to spell out that content is recommended algorithmically. It could also mean that companies have to label algorithmic processing of data (e.g. job applications, access to financial products) more clearly. In some cases, such as Chile’s Bill to regulate AI, robotics and associated technology (Bill 15869-19, proposed by the Congress in 2023 and later combined with an executive proposal), the mechanisms disclosing AI interaction are not specified.
2. Requiring full traceability and auditability of decisions. These proposals, which go further than the first category, require all decisions made by an AI system to be explainable and understandable. In frameworks that categorize AI systems by level of risk, these requirements are often proposed only for high-risk AI uses. Chile’s proposed 2024 “Artificial Intelligence Law” (Bill 16821-19) says in Article 8 that high-risk systems “will be designed and developed with a sufficient level of transparency so that their users and recipients reasonably understand how the system works, in accordance with its intended purpose.” This provision also says that when high-risk systems come on the market, they will be required to use state-of-the-art technical means for interpretability.

3. Creating a right to human review. Many legal proposals, particularly those that focus on auditability or explainability of an AI system’s decisions, further enshrine the right to human review. “AI Bill of Rights” provisions, such as Articles 5-12 of Paraguay’s 2025 Bill that regulates and promotes the creation, development, innovation and and implementation of artificial intelligence systems, regularly include a right to challenge algorithmic decisions, a right to opt out of AI decisions altogether and a right to request human review of any decision. However, the Paraguayan bill does not lay out specific processes for requesting this review, nor does it identify who would conduct it.

4. Reporting requirements and creating public logs and records. Many proposals require developers and providers to register requests for things like licenses and problems like failed audits with the relevant government agency. In many cases, reporting requirements are associated with the creation of public records, so that the information is made fully transparent to the public. For example, Chile’s Bill to regulate AI, robotics and associated technology (Bill 15869-19, proposed by the Congress in 2023 and later combined with an executive proposal) says, “The commission must have a public registry of: [1] The requests for authorization for development, distribution, commercialization or use of AI systems, signaling explicitly if they were authorized or rejected by the Commission. [2] The serious incidents and defective functioning reported by developers, providers and users as well as the judgment that Commission has adopted.”

5. Requiring documentation to be in plain, accessible language. Another fairly common transparency provision is ensuring that developers and providers offer documentation, and requiring that it be usable and accessible. For example, Article 31 (Transparency and Explainability in the use of Information) of Costa Rica’s 2023 Law for the Responsible Promotion of AI (Bill 23919) reads in part, “The information that providers offer in terms of sale of goods and services should be offered in digital, audio, and photographic format and the information should be concise, complete, correct, clear, pertinent, accessible and understandable for users, according to the right to public information.” The provisions we reviewed do not specify which languages must be provided. This is important since hundreds of languages are spoken in Latin America and the Caribbean, and access to materials is far from guaranteed for speakers of most of them.

6. Requiring AI to cite sources. Mexico’s 2024 Bill for National Law that Regulates the Use of Artificial Intelligence would require AI models to cite their sources of information, but there are no specifications.

Impacts on journalism

Requirements that people must know they are interacting with AI systems could be complicated to implement, depending on how “interacting” is interpreted. For example, many journalism producers use algorithms that personalize or filter content, and it’s not clear how best to label such tools. Similarly, if a journalist uses AI tools to transcribe interview audio and then writes a story that quotes from those interviews, would a reader be “interacting” with an AI system?

Laws requiring accessible documentation could make it easier for journalists (who may not have technical expertise) to report on new technologies and to select appropriate tools for professional use. In general, so-called “sunshine laws” that promote transparency in particular sectors have been important to journalists’ ability to access information, and AI is no exception. Accessible documentation could also make it easier for the general public to be better informed about the technology, as research shows that people almost never read terms and conditions because they are too technical.

Some proposed laws focus more on the transparency obligations of deployers rather than developers, which could place a heavy burden on journalism producers.

The Mexican proposal that requires AI to cite its sources will impact journalism, but it is difficult to predict how. This proposal is not compatible with current LLMs, which frequently cite sources that do not support, or even contradict, their results. Such a law could theoretically drive the development of entirely new models, but some experts worry that the proposal is unenforceable because it lacks an understanding of current capabilities of the technology it seeks to regulate. Citing sources with links might potentially drive more traffic to news sites, which would be beneficial, but some research has found that users click on links from AI outputs infrequently.

Data protection and privacy

AI summary: Latin American and Caribbean countries are making data protection and privacy a key guiding principle of their AI policies by banning indiscriminate surveillance; enforcing existing privacy laws; and requiring data minimization, informed consent and periodic reporting. While banning AI for surveillance might make journalists and their sources safer, this benefit is limited by exceptions for government use.

Approaches

Most countries in the region, but not all, have data privacy laws in place that would likely apply to AI, although a detailed examination of these laws is out of scope of this report. Across countries in Latin America and the Caribbean, data protection and privacy are regularly mentioned as key guiding principles in AI regulation. They appear at least once in a document from every country whose bills, strategies or laws we considered. More than half of the documents we reviewed — 43 out of 80 — addressed data protection and privacy.

These principles are operationalized very differently from one place to another, even within this region.

1. Banning indiscriminate data collection and/or surveillance. A number of proposals spell out explicitly that certain practices on data collection and/or surveillance are forbidden. For example, Article 15 in Paraguay’s 2025 “Bill that promotes AI for the social and economic development of the country” says, “The use of AI is prohibited for indiscriminate surveillance, unauthorized monitoring of private life and massive collection of personal data without the explicit consent of the person or without a duly justified judicial order.” Another common proposal is to ban real-time biometric identification, following the EU AI Act. Such proposals do not often apply to government agencies.

2. Highlighting the relevance of existing data privacy protections and frameworks. A number of countries, such as Brazil, have extensive existing data privacy protections that are similar in scope to Europe’s General Data Protection Regulation (and, in some cases, modeled on it). In these contexts, many laws simply include a provision that all privacy protections will be enforced. Brazil’s 2024 proposed Bill on the Regulation of AI Data Centers (Bill 3018/2024) effectively repeats a number of requirements from their data protection law in Article 4: “The operators of data centers must (1) establish clear data governance policies, encompassing collection, storage, processing, sharing and elimination; (2) designate a Data Protection Representative in accordance with the General Data Protection Law (LGPD); (3) conduct impact evaluations on the protection of personal data periodically and any time there are major changes in the processes or technologies used; (4) implement programs of ongoing training for staff on information security and data privacy, with periodic mandatory refresher courses; (5) assure that sensitive data are treated with the highest level of security and confidentiality.”

3. Minimizing and anonymizing data use. Many provisions call for limiting the use of data wherever possible, and to anonymize and pseudonymize data to minimize the risks of data breaches. For example, Article 23 (“Privacy since Design and by Default”) of Ecuador’s proposed 2024 “Organic Law for the Regulation and Promotion of AI in Ecuador” reads, in part: “AI systems will be configured such that, by default, data will not be accessible to an indeterminate number of people without the intervention of the individual. Techniques will be used to minimize, anonymize, pseudonymize, or encrypt data from collection, transmission and storage …”

4. Informed consent and the right to data correction and/or removal. Another commonly used provision requires informed consent for data use, and includes rights to correction and/or removal of personal data from databases, including the datasets used to train various AI tools. For an example, see Articles 13, 16 and 17 of Peru’s 2024 proposed “Law for promotion and regulation of AI in Peru” (Bill 8223/2023).
5. Monitoring and impact reports. To ensure that privacy obligations are taken seriously, many proposals require periodic monitoring, evaluation and reporting on data privacy. A proposed Colombian law, presented to the Congress in May 2025, would empower the Superintendency of Ministry and Commerce to audit, investigate and take preventive measures to protect data privacy.

6. Data sovereignty through infrastructure. While the laws and proposed laws we reviewed did not typically focus on these concerns, national strategies often highlighted the need for infrastructure to guarantee not just data privacy but data sovereignty. For example, the Brazilian Artificial Intelligence Plan (PBIA) 2024-2028 calls for a global Portuguese language model that can support data sovereignty, as well as a robust public data ecosystem in the sovereign cloud.

Impacts on journalism

Banning AI for surveillance is likely to make journalists and their sources safer, although most such provisions have exceptions for government uses. Given that government surveillance of journalists is already widespread in this region — and that state actors are the biggest threat to journalism in Latin America and the Caribbean — these laws may not impact one of the primary safety concerns journalists face.

Bringing privacy expectations and requirements in line with larger national and international frameworks has the benefit of consistency and could simplify compliance. Many countries have existing national privacy regulations that naturally will impact AI, but reviewing them in detail fell outside the scope of this report.

Public information and awareness

AI Summary: Many Latin American and Caribbean countries are promoting public awareness about AI through government-led initiatives or by holding developers responsible for public education. Broad educational efforts about AI would likely help society, but the implications for news organizations are unclear.

Approaches

Many countries (including, at a minimum, Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Jamaica, Panama, Paraguay, Perú, Uruguay and Venezuela) have laws, bills or policies that include provisions to promote public awareness about AI. Out of the 80 documents we reviewed, 38 addressed this issue. 

In general, Latin American and Caribbean policy proposals take one of three approaches to educating the broader public about AI.

1. Providing government-led education. A government agency, often a newly created AI authority, is held responsible for public literacy and education. For example, Venezuela’s 2025 AI Bill charges the Ministries of Education and University Education in Article 33 with incorporating content about AI into “all levels and modalities of the educational system.” Article 34 assigns these ministries as well as the Ministry of Science, Technology and Innovation with developing training and upskilling programs. In addition, Article 35 says “the State” will provide digital education in AI for the general population.
2. Holding developers and deployers of AI tools responsible for public literacy and education. For example, Article 18 of Argentina’s Bill S-2573-S-2025 (proposed in 2024, one of many competing proposals) specifies that “those who are subject to this law must provide clear, understandable and accessible information about responsible use of AI systems, including limitations, risks and necessary precautions. They must promote education and training in responsible use of AI, and promote awareness of associated risks and best practices to avoid errors and minimize potential harms.” Article 2 defines subjects of this law as “any human or legal, public or private person, who develops, researches, implements, commercializes, offers, distributes, imports or uses AI systems in Argentinian territory,” regardless of the location of the server or company.
3. Requiring educational campaigns without specifying who is responsible. For example, Article 17 of Costa Rica’s proposed 2023 “Law for the Regulation of AI in Costa Rica” (one of three bills under discussion) reads simply: “Training and awareness-raising about human rights in the context of AI will be promoted. Training professionals in ethics and human rights in the AI context will be encouraged.”

Impacts on journalism

Widespread educational and workforce development campaigns are likely to benefit societies as a whole. They may also benefit journalism if increasing numbers of journalists are able to participate. Trust in business far outpaces trust in government across Latin America, suggesting that developers and deployers may be important messengers in this context. Government-created educational content may decrease reliance on educational content published by developers, who have a clear conflict of interest, but it is unclear whether most people would trust government content.

While none of the documents we reviewed provides an explicit role for journalism, the ubiquity of proposals for public awareness and literacy campaigns may still have positive implications for the field. Journalists continue to play an important role in explaining new technologies to the public, so journalist access to this information would be of value to technology companies. It is also important that journalists themselves have deep and on-going knowledge about the various facets of AI and technology broadly, as well as access to a wide pool of sources, to cover the material thoroughly and accurately. Unfortunately, there remain gaps in access to information in Spanish and Portuguese, the most widely spoken languages in the region — and the gaps are far wider for the hundreds of Indigenous languages that are not yet meaningfully supported in AI tools.

Conclusion

It remains to be seen whether Latin American and Caribbean countries will succeed in taking a coordinated regional approach to AI regulation. Currently, the countries positioned to be regional leaders — Chile and Brazil — are debating comprehensive legislation that balances incentives for innovation with meaningful regulation designed to protect against harms. These balanced approaches are being debated with a backdrop of global tension, most recently between the BRICS bloc’s explicit commitment to global AI governance and the U.S.’s push for global deregulation. 

Across countries in Latin America and the Caribbean, proposed AI laws address the full range of issues we considered, and they often explore impacts on the information space. It will remain crucial to ensure that policy language in the region safeguards media independence, a plurality of fact-based information sources and a vibrant digital news ecosystem.

The post Latin America and the Caribbean appeared first on Center for News, Technology & Innovation.

In Europe and Central Asia, AI strategies and laws reflect both harm-based and innovation-based approaches. Throughout this region, including within the EU, there are varying levels of established press freedom that will have an impact on how AI laws and policies affect journalism and freedom of expression. The most significant law, the European Union’s Artificial Intelligence Act, is set to be implemented by 27 member states, nine candidate states and potentially three more countries through European Free Trade Association (EFTA) and European Economic Area (EEA) agreements (Norway, Liechtenstein and Iceland are considering adoption; Switzerland is not). The implementation process of this law has been fragmented, and a number of issues need clarification, such as how the AI Act will interact with other existing EU legislation. This law will affect companies far beyond the region due to the large population and market size it will cover (approximately 593.9 million people¹) and the number of legal systems involved. The EU AI Act also has many extraterritoriality requirements, due to a provider or deployer outside the EU being subject to the AI Act if the AI system is used in the EU, whether or not that was the intention. The EU AI Act focuses on classifying different use cases of AI according to risk levels and establishing obligations based on these classifications. Violations of the law are met with hefty fines, as laid out in Article 99, and can range up to 35 million euros, or 7% of worldwide annual turnover, whichever is higher. 

As EU candidate states, Serbia and Ukraine will be required to pass their own legislation that aligns with the EU AI Act. As a consequence, these countries’ strategies are largely related to the EU AI Act; however, they have some important differences, including a greater focus on innovation and public education. 

It is worth noting that in November 2025, the EU Commission proposed the Digital Omnibus Package. These documents propose multiple changes to the EU digital acquis, such as the simplification of the EU AI Act and the General Data Protection Regulation. However, because the publication of this package fell outside of our timeframe, it was not examined in this report. 

The countries in the region that are not required to implement EU law take a variety of approaches, and many have yet to develop strategies or introduce legislation. The United Kingdom’s AI Opportunities Action Plan and Tajikistan’s AI legislation represent a stark divergence from the EU’s AI Act by focusing solely on supporting innovation. Other countries in the region, such as Azerbaijan, Uzbekistan, Kyrgyzstan, Russia and Kazakhstan, have a mixed approach. Switzerland plans on adapting its existing laws and taking a sector-specific approach to AI rather than focusing solely on preventing harm or fostering innovation. 

By the numbers

Of the 11 AI proposals we reviewed in the region, one specifically mentioned journalism; three addressed freedom of speech or expression; two addressed manipulated or synthetic content; three addressed algorithmic discrimination and bias; four addressed intellectual property and copyright; three addressed transparency and accountability; four addressed data protection and privacy; and seven addressed public information and awareness.

Impacts on journalism and a vibrant digital information ecosystem

Laws regarding AI create both opportunities and challenges for journalists. Outside of the newsroom itself, transparency requirements for those developing and deploying AI make it easier for journalists to investigate the technology and the companies behind it. Furthermore, the EU’s transparency rules on generative AI allow journalists to protect their intellectual property by requesting that their copyrighted material be removed from training data. Within a newsroom,  when journalists are deploying AI, whether it be for internal use or outward facing,  transparency requirements can help encourage ethical practices. In addition, transparency requirements, such as labeling requirements, will impact public trust. 

EU policies that do not require AI-generated text to be labeled when it has been reviewed by a human may inadvertently increase confusion because both malicious actors using AI to spread disinformation and legitimate journalists who use AI but edit the output could avoid disclosing AI use. This differs from stricter rules in Uzbekistan, where all AI-generated or manipulated content must be labelled. Under both systems, however, the level of trust between journalists and the public will likely be affected. Furthermore, the EU’s law does not explicitly ban biometric surveillance in emergencies, which could put journalists and their sources at risk. Outside of the EU, journalists face concerns over copyright since many laws and strategies provide little to no protection for their work being used to train AI models without consent.

Freedom of speech and expression

AI summary: The EU’s Digital Services Act and AI Act work together to protect freedom of expression by banning manipulative AI and requiring transparency from social media platforms, while other countries in the region mention these rights but do not have specific action plans. The AI Act has an exception allowing for biometric surveillance, which could put journalists and their sources at risk.

Approaches

Some freedom of speech and expression concerns will be addressed by the intersection of the Digital Services Act and the AI Act in the EU, especially when it comes to AI use on social media platforms. Countries outside the EU mention freedom of expression or fundamental rights without citing actionable steps. 

1. The overlap of the Digital Services Act (DSA) and AI Act will impact freedom of expression in the EU. The AI Act itself is not specifically designed to protect the freedom of expression, other than its bans on manipulative or exploitative AI systems that distort human behavior and impair decision-making. As stated in the Act’s preamble, the DSA and the AI Act will work together when it comes to AI systems embedded in online platforms. Algorithm recommender systems are not explicitly labeled as “high risk” under the AI Act unless they are deemed to have electoral influence. However, this concept has been critiqued for its vagueness and likelihood to include all digital platforms’ algorithmic recommender systems. Compared to the AI Act, the DSA has more stringent transparency and user control requirements. When it comes to protecting freedom of speech, the DSA has more enforceable protections than the AI Act, which has high evidentiary burdens when it comes to fundamental freedom violations. Other countries, such as Ukraine, plan to align with the AI Act when it comes to protecting fundamental rights.

2. Mentioning freedom of expression without including actionable steps. Serbia’s AI strategy mentions freedom of expression, emphasizes how AI will impact the information environment as a whole and discusses potential impacts on other fundamental rights such as protection from discrimination. However, it does not have actionable steps in these areas outside of some suggestions, such as educating the public and media professionals. Countries such as Kazakhstan, Kyrgyzstan, Russia and Switzerland did not mention freedom of speech, expression or information but did mention fundamental rights as a whole. 

Impacts on journalism

The AI Act does not mention the right to reliable information or quality journalism when discussing fundamental rights. Although democracy and access to information are typically considered to go hand in hand, the lack of specific mention in such a comprehensive law may leave fundamental rights vulnerable. The DSA, AI Act and the European Media Freedom Act overlap when it comes to moderating content news organizations share on online platforms. This overlap might have impacts on the reach of a news outlet’s content, but it is not yet clear if this will be positive or negative. News organizations’ own recommender systems on their websites would not be impacted by the DSA.

The EU AI Act also leaves open the potential for member states to implement some prohibited kinds of AI, such as biometric surveillance in cases of emergency. This exception may leave journalists and their sources vulnerable to surveillance by national authorities and the police.

Manipulated or synthetic content

AI summary: Policies throughout Europe and Central Asia have different rules for labeling AI-created content, with the EU requiring watermarks on deepfakes and Uzbekistan requiring that all manipulated content, including text, be labeled without exception. The differing rules on labeling AI-generated content can create challenges for journalism, potentially allowing false information to be shared by actors pretending to be real news sources, which could lead to a loss of public trust.

Approaches

While manipulated content is an important part of the EU AI Act and the Uzbekistan draft law, it is not addressed in other documents we reviewed. These two documents share a common approach. 

1. Requiring manipulated content to be watermarked and machine-readable. Article 50 of the EU AI Act requires deployers of AI systems that generate deepfakes to ensure that their outputs are watermarked in a machine-readable format. Image, audio and video content must always be labeled, with exceptions only for law enforcement purposes. In cases where an AI system generates or manipulates text intended to inform the public on matters of public interest, deployers must disclose that the content was artificially generated unless it has been subject to human review or editorial oversight. In Uzbekistan’s draft law, manipulated content must be labeled, with no exceptions, through watermarks or other machine-readable formats. 

Impacts on journalism

Policies on deepfakes and other synthetic content carry significant implications for journalism. In both the EU and Uzbekistan, further clarification on the definition of AI-generated material would benefit journalists using this technology. For example, it remains unclear whether an image or video edited with AI should be classified as AI-generated content and therefore subject to labeling requirements. Generally, in the EU, AI-generated content such as images, videos, audio files and text must be labeled as such; however, if a newsroom reviews textual content and assumes editorial responsibility, labeling is not required, unless the content could mislead the public. As a result, in many cases, text generated through AI used by newsrooms will not need to be explicitly marked as AI-generated, whereas textual material not subject to review must carry a clear, machine-readable and visible label. (Malign actors posing as legitimate news sources to spread disinformation could also take advantage of the EU’s rules by claiming human review of AI-generated content, even when no such review took place.) However, AI-manipulated images, video and audio must always be watermarked and labelled as artificially generated or manipulated in a machine-readable format. These kinds of labels, however, might create confusion and short-term mistrust among audiences, according to some studies.

In contrast, Uzbekistan mandates labeling for all AI-generated content, regardless of editorial oversight, which demonstrates the importance of public understanding of manipulated content but could actually undermine trust if labeling is poorly explained or understood.

Algorithmic discrimination and bias 

AI summary: Different laws across the region, especially the EU AI Act and General Data Protection Regulation (GDPR), are working to reduce algorithmic discrimination, while some countries like Kyrgyzstan and Switzerland mention the concern but do not offer specific actions. The use of AI by journalists for things like creating content or analyzing data will require them to be aware of and avoid biased outputs, especially when dealing with personal information about different groups.

Approaches

Algorithmic discrimination is a key concern addressed in numerous laws and strategies throughout the region, and is a major point in the EU AI Act. 

1. Requiring the prevention, detection, and mitigation of discriminatory outcomes on high-risk AI systems. Under the EU AI Act, developers must ensure that their training, validation and testing data meet quality criteria to help reduce bias. They can do this by adhering to Article 10(5) of the AI Act, which allows providers of high risk AI systems to process special categories of personal data under strict conditions to identify, assess and mitigate bias/discrimination, which would normally not be allowed under the General Data Protection Regulation (GDPR). 

2. Banning some AI applications. In Article 5 of the EU AI Act, some AI applications are banned because they are seen as inherently discriminatory or carrying high risk of discrimination. These include social scoring and the use of biometric data to classify people in ways that would violate non-discrimination or equality principles. 

3. Clarifying the applicability of existing laws. Switzerland, for example, is clarifying how AI relates to labor laws.

4. Mentioning these concerns without actionable points. Multiple countries raise concerns about algorithmic discrimination without specifying a course of action. For example, Kyrgyzstan’s law mentions the concept and considers it with the idea of data protection, but it does not present actions to address bias or discrimination. 

Impacts on journalism

Throughout the region, the concept of algorithmic discrimination will be important when journalists select AI systems to use for creating content, completing administrative needs and more. The EU’s call to prevent, detect and mitigate discriminatory outcomes in high-risk AI systems has clear implications for journalism, especially for newsrooms developing or deploying AI tools for reporting, content moderation or audience analytics. This could include using AI to analyze datasets for reporting or using audience data with personally identifying or other sensitive information. Journalists using or building AI tools will need to understand these obligations and make sure they and their newsrooms comply. 

Outside of the EU, the fragmented landscape will require journalists to balance the ethical use of AI with compliance across jurisdictions, and it will require them to remain aware of how bias-related safeguards may limit or shape their use of different tools. 

Intellectual property and copyright

AI summary: In the EU, the AI Act and the Directive on Copyright work together to give copyright holders the ability to withhold their content from AI training, while other countries, like the United Kingdom, Switzerland and Kyrgyzstan, have less specific plans and are starting to update their existing laws. The U.K. government held a consultation in December 2024 on introducing a text and data mining (TDM) exception for AI training that will likely be similar to the EU model. The impact of these different laws is a complex legal situation where journalists’ work in countries outside the EU could be used to train AI without their permission or payment, while within the EU the opt-out system is seen as a positive step for journalism but may not be fully effective due to vague rules.

Approaches

The EU AI Act overlaps with the EU’s Directive on Copyright, a law harmonizing copyright rules across EU member states to reflect how works are created, shared and consumed online. The intersection of these two laws will drive most of the copyright protections within the EU, while other countries in the region have opted to change their existing copyright laws without offering many details. 

1. Giving copyright holders the ability to withhold their content from AI training. Under the EU Directive on Copyright in the Digital Single Market, text and data mining of copyrighted materials is allowed for research purposes, but rights holders can opt out for commercial purposes, including AI model training. “General-purpose” AI models have specific transparency requirements under the AI Act, including maintaining and sharing technical documentation with downstream providers, respecting copyright law and publicly disclosing a detailed summary of training data. 

2. Requiring transparency about training data for generative AI. The EU AI Act’s Article 53 requires that sufficient detail about an AI system’s training data be publicly available so that copyright holders are able to opt out; however, AI companies do not have to publish their full datasets. Further details on transparency requirements have now been published by the EU in the Code of Practice for General Purpose AI, though this fell out of our research period. 

3. Changing existing copyright laws. In Kyrgyzstan, copyright issues are mentioned regarding the exchange of data with developers. Switzerland plans to update its existing federal copyright act, and while the United Kingdom’s action plan states that the country will “reform the U.K. text and data mining regime so that it is at least as competitive as the EU,” the U.K.’s approach to copyright issues is still being considered. However, these three AI strategies or laws do not go into specifics. 

Impacts on journalism

The impacts of copyright provisions exist within a complicated and fragmented legal landscape on the issue. Not only are laws significantly different country to country, but many do not account for the digital environment nor have begun to grapple with the impacts of AI. 

In the EU, the Directive on Copyright in the Digital Single Market and the AI Act will address this issue. For journalists in the EU, the AI Act’s existing opt-out mechanism for copyrighted material has raised concerns. While the Act includes a provision allowing rights holders to opt out of having their content used for AI training, critics argue that vague implementation guidelines risk making the safeguard ineffective. On the other hand, organizations such as Reporters without Borders view the situation as a chance to push for greater transparency requirements to make it easier for authors and other rights holders to protect their intellectual property. More detailed requirements for the disclosure of generative AI systems training data were released in July 2025, including the Model Documentation Form and Public Summary Template. 

Outside of the EU, copyright legislation and protections become much more complicated in the region. The strategies and legislation from outside of the EU have less of a focus on copyright, potentially posing a problem for journalists as their intellectual property could be used to train these AI systems without permission, compensation or recognition. 

Transparency and accountability

AI summary: The EU AI Act has strict rules for transparency and accountability for high-risk and generative AI systems, which include requiring detailed documentation about the system and informing users when they are interacting with an AI. While some other countries in Europe and Central Asia mention the need for transparency in their proposals, they have not yet provided detailed or actionable plans. These transparency rules can help journalists investigate AI technology, allow news organizations to protect their work from being used as training data and build public trust in the use of AI in the media.

Approaches

Outside of the EU AI Act, there is limited mention of transparency and accountability for AI systems in Europe and Central Asia. 

1. Requiring clear user instructions in accessible language. Transparency for users of high-risk AI systems is required under the EU AI Act’s Article 13. Users must be provided with clear instructions on how to use the system, as well as clear information on system capabilities, limitations, known risks, intended purpose and necessary human oversight measures. 

2. Informing the public when they are interacting with an AI system. Under the EU’s AI Act Article 50, the public must be informed when they are interacting with an AI system, AI-generated content must be labeled clearly and in a machine-readable format (notwithstanding the exceptions covered under Manipulated Content) and users must know if an AI system is making decisions that impact them, especially for biometric recognition or content moderation.

3. Requiring high-risk systems to provide details about training data. Systems classified as high risk under the EU’s AI Act must be transparent about their training data under Article 10. The data must be relevant, representative and as free of errors as possible. Providers must document how the data was collected and processed, and how they plan to mitigate bias or discrimination. 

4. Instituting audits of documentation to ensure it is accurate and up-to-date. Under the EU AI Act Article 10 and Article 11, providers are required to maintain technical documentation and record keeping of AI systems’ datasets, design, development, testing and risk management, all of which must be available to authorities for inspection.

5. Emphasizing transparency as a value without actionable steps. The Uzbekistan draft law refers to a “proactive approach to transparency and ethics” but does not provide details on the approach. 

6. Reforming existing laws. In Switzerland the government plans to address transparency concerns through the reform of existing labor laws and sector-specific regulations. 

Impacts on journalism

Transparency obligations have important implications for journalism. Broadly, increasing transparency can make it easier for reporters to investigate AI technology and the companies that provide it. Additionally, the EU’s transparency requirements for high-risk and generative AI systems will allow journalists to protect their copyrighted material by allowing them to request that material be removed from AI training data.

When it comes to newsrooms using AI, they must ensure that the systems they implement are ethically transparent and compliant with national laws. In the EU, this could involve several strict transparency requirements if the system is determined to be high risk. Additionally, if a news outlet uses AI-generated audio, photos or video, it must be disclosed as such. News outlets must also label AI-generated text unless it is subject to human review and someone takes editorial responsibility. The requirement to label AI-generated content, with exceptions for text, will likely influence the public’s level of trust in the media.

Data protection and privacy

AI summary: The EU AI Act and the EU’s General Data Protection Regulation (GDPR) work together to address data protection and privacy concerns for AI systems, especially for personal data, while countries outside the EU, such as Kyrgyzstan, Russia and Uzbekistan, have focused on specific aspects of data protection like anonymization or alignment with global trends. These rules could require news organizations to use human oversight for AI systems that make important decisions and could make it very difficult to handle requests from people who want their personal data deleted from an AI’s training dataset when a journalist is using or deploying an AI system.

Approaches

Data protection and privacy concerns will be addressed by the EU AI Act’s overlap with the GDPR. Countries outside the EU focus on specific aspects of data privacy. 

1. Protecting individual rights. GDPR Article 22 provides individuals with rights against fully automated decisions that have significant impact. Additionally, under the GDPR, individuals have the right to be forgotten or have their information deleted. Due to the way data is collected and processed by AI systems, this right could be particularly hard to protect. National Data Protection Authorities and National Competent Authorities (Market Surveillance Authorities and Notifying Authorities) will work together when it comes to data protection in each member state. 

2. Requiring data anonymization and deletion. Such requirements are found in Kyrgyzstan’s law, which mandates data handling requirements for the design and use of AI systems. This includes requirements for the quality of training, validation and control samples. Russia’s Data Protection Law has guidelines on data anonymization that will impact the use of personal data in AI use and development by requiring additional levels of transparency.

3. Deprioritizing privacy and data protection in favor of innovation. The United Kingdom’s strategy states, “Prioritisation should consider the potential economic and social value of the data, as well as public trust, national security, privacy, ethics, and data protection considerations.” This is an example of their pro-innovation approach that contrasts with the harm-based approach taken by the EU. 

Impacts on journalism

GDPR’s Article 22 might require news organizations to implement human oversight for the use of AI for personalized content delivery, audience segmentation or moderation. If AI systems use personal data for training, including data from past news stories or any other source, deletion requests will become complex for the deployer, which could be the newsroom or the journalist in some circumstances, but could be an AI company that scraped news content. If the journalist is the end user of the AI system, it would be the responsibility of the AI deployer to delete the content. 

Public information and awareness

AI summary: Laws in the region use various methods to teach the public about AI. Azerbaijan and Russia, for example, focus on public education programs, while Serbia focuses on training journalists, and others like Tajikistan and the United Kingdom focus on integrating AI into school curricula. The laws and strategies in the region do not currently involve journalists as key partners in public AI education, which is a missed chance to broaden the reach and impact of information about AI.

Approaches

Laws in this region take varied approaches to educating the public on AI. 

1. Establishing public sector education programs. Azerbaijan and Russia each have a law and strategy plan to establish education programs both within and outside of traditional education institutions. Additionally, Azerbaijan has set its sights on becoming a regional hub for AI education and plans to accomplish this by creating an AI academy for public sector education with a focus on workforce development. 

2. Training journalists on AI. This is a unique element in Serbia’s strategy. Serbia plans to train journalists to use AI ethically, enable greater accessibility for audiences with disabilities, and emphasize protection of freedom of expression and protection from hate speech. The country plans on organizing seminars on information security, AI and the use of big data for media professionals.

3. Incorporating education about AI into secondary and higher education. Plans to incorporate education about AI into secondary and higher education are an important element of AI strategies in Tajikistan and the United Kingdom. Tajikistan’s strategy includes provisions to develop AI clubs in education centers throughout the country, implement a “Fundamentals of Artificial Intelligence” class in vocational education systems and create an AI module for senior classes of secondary education institutions.

4. Educating the public on AI use in healthcare. Kyrgyzstan’s law emphasizes debunking myths around the use of AI in healthcare and demonstrating its practical implementation.

5. Supporting public education without specific guidance. The EU AI Act mostly leaves it to member state discretion when it comes to increasing public education and awareness on AI. It emphasizes the importance of promoting AI literacy and encourages member states to support training and education initiatives. It also emphasizes that the AI Office and national competent authorities are encouraged to promote AI education and awareness, including by developing training materials and supporting public campaigns. Similarly, Serbia’s AI Strategy mentions education on AI and media literacy for the public; however, it does not provide specific guidance. 

Impacts on journalism

On paper, Serbia’s strategy offers a strong example of including training programs to help journalists adapt to AI. However, Serbia’s media is in a position of state capture, which raises doubts about this provision’s role in facilitating a positive environment for media as Serbia integrates AI. In other countries, expanding education on AI within the public sector could further benefit the media because future journalists would gain a deeper understanding of these technologies, and media consumers would be more informed, potentially strengthening public trust in journalism. 

Conclusion

As AI legislation continues to evolve across Europe and Central Asia, the successful implementation of the EU AI Act will be a crucial factor in determining the future of journalism in the AI era. This landmark law is positioned to shape the region’s approach to artificial intelligence as well as global standards, as many countries move to align their frameworks with its provisions — including countries outside the EU. As the implementation process begins, other countries in the region are expected to develop or revise their own AI strategies and legal frameworks. 

One significant area of concern is how the AI Act will be enforced in countries, both within and outside the EU, that do not adhere to democratic norms. The effectiveness and integrity of its implementation in these contexts will be crucial to avoid negative impacts on fundamental rights.

Current laws and strategies in the region do not explicitly address journalism or journalists. All the same, their indirect effects will be significant on the field. This is particularly true in countries within the EU, like Hungary, Romania and Bulgaria, and candidate countries such as Serbia and Bosnia and Herzegovina, where press freedom is threatened. Ongoing monitoring will ensure that AI governance does not further undermine media independence.

1. The population of Türkiye is included in this number, but its accession process and negotiations have been frozen. ︎

The post Europe and Central Asia appeared first on Center for News, Technology & Innovation.

AI regulation in the Middle East and North Africa is largely at the nascent stage. AI readiness varies widely across the region, with the United Arab Emirates ranking 13th in the world for AI readiness, while Yemen comes in last place in the world, according to Oxford Insights’ 2024 Government AI Readiness Index. 

Already, though, several countries have developed national AI strategies or policies, which set standards for the role of AI across sectors. These strategies take a mix of innovation-based and harm-based approaches, demonstrating that AI is not always viewed the same way across the region. Regardless, countries in the region have found common ground on AI, with the Gulf Cooperation Council (of which Bahrain, Kuwait, Oman, Qatar, Saudi Arabia and the United Arab Emirates are members) endorsing the Bahrain-drafted Guiding Manual on the Ethics of Artificial Intelligence Use. Some North African countries have also highlighted their intention to embrace the African Union’s Continental AI Strategy and to cooperate with other countries on the continent. 

While countries have derived different definitions of AI — when one is included in these proposals at all — most associate AI with the “simulation of human intelligence.” Libya’s National AI Policy, for example, notes that “​​Artificial intelligence is a branch of computer science that simulates human cognitive abilities through intelligent machines,” while Qatar’s Guidelines for Secure Adoption and Usage of Artificial Intelligence state that AI is “is designed to carry out any tasks associated with human intelligence, in a manner that mimics the human mind with a certain level of autonomy.” Policies in Oman, Jordan, Saudi Arabia and the United Arab Emirates (UAE) similarly define AI as attempting to perform functions that are generally associated with human cognitive abilities. This language highlights that AI is not omnipotent; it attempts to mimic the outputs of the human brain, but it ultimately does not work the same way as human reasoning and cannot be expected to mimic the human mind without fail. As such, many countries include requirements for human oversight over AI systems in their strategies, meaning that humans should review these technologies and make the final decisions, instead of relying on AI to tell humans what to do on any given topic. This is especially important for journalists: Because AI is fallible, and because good reporting must be based on verifiable facts, journalists should never rely solely on AI technology. Likewise, they should fact-check any AI-generated information they receive from sources. 

As of June 2025, two pieces of AI legislation have been passed in the region: the government of Abu Dhabi’s Law No. 3 of 2024, which established the Artificial Intelligence and Advanced Technology Council (AIATC) to regulate projects, investments and research related to artificial intelligence and advanced technology in the emirate of Abu Dhabi; and Qatar’s AI Guideline to Regulate the Use of AI by Qatar Central Bank Licensed Entities, which regulates the use of AI in the country’s financial sector. Neither of these laws have clear implications for journalism yet; although, because the AIATAC is charged with guiding policies and strategies for AI in Abu Dhabi, it is possible that its future rulings will impact journalism. The governments of Bahrain and Egypt are developing comprehensive AI laws that will impact journalism when entered into force — Bahrain’s draft AI Regulation Law was unanimously approved by the Shura Council in 2024 and is awaiting parliament’s approval, and Egypt is reportedly in the final stages of developing a set of laws to regulate AI and dataflows. 

By the numbers

Of the 19 AI proposals we reviewed in the region, two specifically mentioned journalism; none addressed freedom of speech or expression; seven addressed manipulated or synthetic content; 13 addressed algorithmic discrimination and bias; six addressed intellectual property and copyright; 15 addressed transparency and accountability; 14 addressed data protection and privacy; and six addressed public information and awareness. 

Impacts on journalism and a vibrant digital information ecosystem

Each of these seven policy areas has the potential to significantly impact journalism and the information space across the region. AI awareness, literacy and education campaigns could provide journalists with the knowledge they need to cover AI, but journalists will have to ensure their coverage remains independent. Attempts to counter some of the risks associated with AI-manipulated content could help prevent online violence against journalists, but vague wording in these laws could also be used to penalize journalists for reporting on or using AI. Journalists and newsrooms will have to strike a balance between using AI to personalize content and ensuring their AI tools avoid algorithmic discrimination and bias. Some proposals on intellectual property and copyright would benefit journalists by requiring AI companies to remunerate them for the use of their content. Transparency efforts are generally beneficial for journalism, as they can improve journalists’ understanding of how AI technology is being used and how it is arriving at its conclusions, thus enabling journalists to better report on the technology and the entities using it. Journalists and new organizations will be required to build data protection and privacy into their AI systems, which could protect journalists and their sources. Proposals that prohibit the use of AI for surveillance would also be a boon to journalists in the region. Finally, the lack of inclusion of freedom of speech and expression in the reviewed AI policies mirrors the lack of press freedom that journalists experience in the region.

Freedom of speech and expression

AI summary: AI proposals in the Middle East and North Africa do not mention freedom of speech or expression, reflecting the region’s overall low ranking in press freedom and the ongoing threats journalists face.

Approaches

None of the reviewed policies, strategies or laws specifically mention freedom of speech or expression. 

Impacts on journalism

The lack of inclusion of freedom of speech and expression in the reviewed AI policies does not necessarily signify that countries in the Middle East and North Africa do or do not support this right. In fact, with the exception of Morocco, the North African countries covered in this section could choose to follow the African Commission’s 2019 Declaration of Principles on Freedom of Expression and Access to Information in Africa. The Declaration is not enforceable, but it does establish a normative framework that member states of the African Union should seek to uphold. 

Nonetheless, according to Article 19’s 2025 Global Expression Report, the Middle East and North Africa has the lowest expression score of any region, with not a single country ranking as “open.” There is a close connection between freedom of speech and expression and freedom of the press — freedom of speech creates the environment in which freedom of press can exist. The lack of consideration for freedom of speech paired with the low rankings across the region may signify a negative environment for press conditions. Reporters Without Borders’ 2025 World Press Freedom Index reinforces this. According to the index, the existence of the free press in the Middle East and North Africa continues to be hampered by violence, repressive legal systems and political control, even though freedom of expression is constitutionally enshrined in some countries, such as Jordan and Israel. 

Manipulated or synthetic content

AI summary: Proposals from several countries in the Middle East and North Africa aim to address the risks of AI-generated disinformation by prohibiting certain uses of AI, setting standards for AI development, or using AI to help detect false content. While these efforts could help protect journalists from manipulated content and improve the quality of the information space, some of the vague rules might lead journalists to censor themselves.

Approaches

Several proposals throughout the region note that AI can exacerbate the risk of mis- and disinformation, including through the generation of deepfakes, but only a few lay out strategies for addressing manipulated content. 

1. Prohibiting some uses of AI. Bahrain’s draft AI Regulation Law strictly prohibits the use of AI for the “the fabrication or installation of personal photos that harm an individual’s reputation, honor, or dignity; manipulation or falsification of speeches, statements, or official communications; manipulation of any textual, audio, or visual content of individuals without their explicit consent; manipulation or falsification of personal, health, or professional data; [or] any other cases determined by a decision issued by the Minister, based on a proposal from the Artificial Intelligence Unit.” Violators face regulatory fines or up to three years in jail. 
2. Providing mitigation measures. Saudi Arabia’s Generative AI Guidelines for [the] Public outline mitigation measures for both “deepfakes and misrepresentation” and “misinformation and hallucinations.” The former include watermarking, output verification and enhanced digital literacy. The latter include content verification and citation, content labeling and fact-checking. 
3. Using AI to counter mis- and disinformation. Egypt’s National AI Strategy, for example, notes that AI can be used to help detect “fake news” and asserts that the government can use AI to combat mis-, dis- and malinformation (MDM). Libya’s National Artificial Intelligence Policy also states that the government can use AI to counter MDM. Neither proposal explains how the government would use AI to do so. 
4. Setting value-based standards. Saudi Arabia’s AI Ethics Principles include instructions for AI developers: “Predictive models should not be designed to deceive, manipulate, or condition behavior that is not meant to empower, aid, or augment human skills but should adopt a more human-centric design approach that allows for human choice and determination.” The principles lay out how developers can accomplish this at every stage of the AI lifecycle, including by properly acquiring and processing data and by conducting periodic assessments. 
5. Noting the risks. For example, Israel’s Policy on Artificial Intelligence Regulations and Ethics states, “The proliferation of fake news and disinformation, with attendant risks to democratic governance, harms to fundamental rights and freedoms, wide-scale consumer manipulation, and the like. A discussion about the appropriate scope of disclosure for AI systems must take these broader concerns into account.” Saudi Arabia’s Generative AI Guidelines for Government similarly note that government users of AI tools must take into account the risks arising from their use, such as misinformation and deepfakes.

Impacts on journalism

Bahrain’s prohibition on some forms of manipulated content may help prevent forms of technology-facilitated violence against journalists, such as deepfake nudes, and prevent their likenesses from being used in content they are not involved with. This would help protect journalists from violence, which could in turn enable journalists to safely work in the profession. On the other hand, the broad wording of the draft law may pose problems for reporters. It is not clear, for example, whether a reporter could be penalized for using AI to make minor edits to an image, nor is it clear how “damage” to an individual’s reputation is decided. Due to the lack of clarity and the threat of punishment, journalists may self-censor. 

Attempts to use AI to counter mis- and disinformation are generally positive for the information space, and some early examples have shown this strategy can be effective; however, there must be clear boundaries between countering verified disinformation and censoring unwanted narratives.

Attempts to set standards and mitigation measures can provide news organizations and others with good practices to follow as they implement AI in their work, which will help prevent the further degradation of the information space. Because Saudi Arabia’s guidelines are not legally binding, however, this may lead to a split between organizations that choose to adopt these measures and organizations that do not; such a split could inadvertently further corrupt the information space by allowing bad actors, such as “disinformation for hire” firms, to hide their use of manipulated content, while more legitimate outlets that choose to adopt the measures might be viewed with distrust for admitting they use AI. 

Countries who note the risks posed by AI-manipulated and synthetic content have taken an important first step, but they will need to further consider how best to address the issue to truly have a positive impact on journalism and the information space. 

Algorithmic discrimination and bias

AI summary: In an effort to prevent bias and discrimination, several countries in the Middle East and North Africa are creating rules for how AI should be designed, which includes requiring audits, using diverse datasets and fining those who create biased systems. While these rules could help journalists make sure their reporting is fair, they may also make it hard for newsrooms to use AI to personalize content.

Approaches

The Gulf Cooperation Council’s Guiding Manual on the Ethics of Artificial Intelligence Use in Member States lays out recommendations member states can take to prevent bias in AI, such as auditing datasets, establishing oversight processes, addressing the digital divide, engaging diverse stakeholders and ensuring diversity in access to AI systems. Both members of the Gulf Cooperation Council and non-members in the region have taken similar approaches to address algorithmic discrimination and bias in their AI proposals: 

1. Requiring algorithmic audits. Libya’s National AI Policy, Oman’s Public Policy for the Safe and Ethical Use of Artificial Intelligence Systems and the UAE’s AI Ethics Guide require developers to conduct algorithmic audits to prevent bias.
2. Using diverse datasets. Saudi Arabia’s AI Ethics Principles, for example, call for AI systems to be trained on data that is “cleansed from bias” and “representative of affected minority groups” — and it lays out steps developers can take to ensure tools are unbiased at every stage of the AI lifecycle. Saudi Arabia’s Guidelines for Government further note the importance of understanding where the data comes from to prevent bias. The UAE’s AI Ethics Guide also lays out steps developers can take to mitigate and disclose the biases inherent in datasets, including hiring people from diverse backgrounds to work at every stage of the AI lifecycle. The UAE’s AI Adoption Guideline in Government Services also notes that the data used to develop AI models across government services “should be representative, otherwise AI decisions will turn out to be biased.” 
3. Improving users’ understanding of bias. Saudi Arabia’s Generative AI Guidelines for Government and Generative AI Guidelines for Public both feature components calling for an increased understanding of bias for users. The former implores government developers to “enhance users’ understanding and awareness of bias, the importance of diversity and inclusion, anti-racism, values and ethics, as this knowledge will contribute to improving their ability to identify biased content.” The latter calls on developers and end users to enhance their knowledge of bias to improve their ability to recognize biased content. 
4. Imposing fines. Article 25 of Bahrain’s draft AI Regulation Law imposes fines on those who “design programs or systems that result in discrimination between persons who enjoy equal legal status.”

Impacts on journalism

As journalists and newsrooms develop and deploy AI systems, they will have to ensure their AI tools follow best practices to avoid algorithmic bias and discrimination both to ensure their reporting remains objective and to ensure they do not fall subject to fines or other penalties. Even when using diverse datasets and with algorithmic audits in place, however, journalists should always ensure that a human confirms the accuracy of any AI-generated content. 

On the other hand, national regulations can pose a challenge to newsrooms’ efforts to use AI to personalize content using more sensitive data. Bahrain’s draft law, for example, could potentially be used to penalize news organizations, depending on how “discrimination” is defined in the law and how the law is applied to the media sector. News outlets will have to walk a fine line between personalization and bias mitigation. To explain how they do this, newsrooms could consider publishing their own AI ethics policies, which should outline how they are using AI technology, how they are mitigating biases and discrimination, and how they implement human oversight of all AI-produced content. 

Intellectual property and copyright

AI summary: Countries in the Middle East and North Africa are taking different approaches to handle intellectual property and copyright for AI, such as requiring licenses for content, creating new patent systems or updating current laws. These rules could help journalists by requiring AI developers to pay journalists for the use of their content to train AI models, but it is still important that new policies properly protect journalistic work.

Approaches

Intellectual property rights and copyright are covered in a handful of AI policies throughout the Middle East and North Africa, each one taking a slightly different approach. 

1. Mitigating copyright infringement. Saudi Arabia’s Generative AI Guidelines for Public lays out mitigation measures for intellectual property infringement, calling on generative AI developers to obtain licenses for any intellectual property they use in training data, which would prevent “indiscriminate data scraping.” It also requires AI developers to establish compensation mechanisms to pay creators whose intellectual property has been used to train AI. 
2. Developing a patent granting system. Egypt’s National AI Strategy outlines an initiative for an AI patent granting system, which would raise awareness of intellectual property protection and develop AI-specific patent categories for “AI inventions.” This part of the strategy is focused on enabling innovation, but it is not yet clear what the patent categories would be, nor is it clear who would own AI-generated content. 
3. Applying existing laws to AI. Jordan’s National Ethics Charter for AI explains that any data used to train AI models should respect intellectual property laws. 
4. Updating existing legislation. Oman’s National Program for AI and Advanced Digital Technologies, for example, suggests the government update the Copyright and Neighboring Rights Protection Act to account for generative AI, but it does not detail what updates should be made.

Impacts on journalism

The use of news articles and other journalistic content to train AI models for commercial use raises important copyright questions. Proposals that would mitigate this infringement by requiring AI companies to obtain licenses to use content and/or remunerate the content creators could provide an important lifeline for journalists and news organizations — many of whom report losing traffic to their original news articles due to AI. 

A patent granting system is likely to benefit AI developers who create “AI inventions.” In some situations, it could also benefit journalists who use AI to produce original materials. Further clarity on this concept will be essential, though, and policymakers will need to include specific considerations for journalism and other forms of media.

As countries determine how best to update or apply their existing copyright laws, policymakers should also consider how best to protect journalistic content from copyright violations in the AI era. 

Transparency and accountability

AI summary: Most of the AI policies in the Middle East and North Africa require transparency and accountability, using methods like audits, user notices and rules for who is responsible for AI-caused harm. These efforts can help journalists better understand and report on AI, but they also mean newsrooms may be held responsible for any damage caused by the AI tools they use in their work.

Approaches

Nearly all reviewed documents in the region included provisions for transparency and accountability, but they take a wide variety of approaches.

1. Requiring audits. The UAE’s AI Ethics Guide, Egypt’s National AI Strategy and Qatar’s Guidelines for Secure Adoption and Usage of Artificial Intelligence all stress the importance of conducting audits to improve transparency. 
2. Building in appeals procedures. To ensure AI systems are transparent and accountable, the UAE’s AI Ethics Guide notes that AI systems should have built-in appeals procedures that would allow users to challenge “significant decisions.” The guide further suggests that “AI operators and AI developers should consider designating individuals to be responsible for investigating and rectifying the cause of loss or damage arising from the deployment of AI systems.”
3. Informing users they are interacting with AI through labels or other notices. Israel’s Policy on Artificial Intelligence Regulations and Ethics states, “To the extent possible and in appropriate cases, individuals should be: (1) informed that they are interacting with an AI system, (2) notified if an AI system is being used to make recommendations or decisions involving them, and (3) provided with an understandable explanation of an AI-based recommendation or decision involving them,” echoing the UAE’s AI Ethics Guide. Saudi Arabia’s Generative AI Guidelines for Government and for the Public are slightly more specific, requiring government entities to communicate when using generative AI to interact with the public and to use watermarks to help consumers identify AI-generated content. Oman’s Public Policy for the Safe and Ethical Use of Artificial Intelligence Systems calls on developers to provide mechanisms to identify AI-generated content through labeling or explanatory notices to prevent misuse. 
4. Determining liability. The UAE’s AI Ethics Guide, for example, states that “accountability for the outcomes of an AI system lies not with the system itself but is apportioned between those who design, develop and deploy it.” Bahrain’s draft AI Regulation Law notes that anyone who suffers damage due to an AI system has the right to demand compensation from the programmer, processor or developer. Saudi Arabia’s AI Ethics Principles hold designers, vendors, procurers, developers and assessors, as well as the technology itself, ethically responsible and liable for an AI system’s decisions and actions, noting that these entities should be identifiable and take accountability for any damages. 
5. Calling for decisions to be both explainable and replicable. Oman’s Public Policy for the Safe and Ethical Use of Artificial Intelligence Systems requires AI developers to provide documentation explaining an AI’s decision-making logic and data analysis process; to ensure that AI systems are capable of providing understandable explanations of the decisions they make; and to ensure an AI documents all processes to enable future verification and analysis. Saudi Arabia’s Generative AI Guidelines for Government also state that transparency is needed across the design and implementation of an AI system, as well as in the product and its justifications of an outcome. AI systems’ decisions should be traceable, according to the guidelines, and companies should include an information section on their platforms. 

Impacts on journalism

Transparency efforts are generally beneficial for journalism, as they can improve journalists’ understanding of how AI technology is being used and how it arrives at its conclusions, thus enabling journalists to better report on the technology and the entities using it. 

These provisions will also impact the way journalists and newsrooms use and deploy AI. According to these policies, if journalists use AI in their reporting, the AI’s decisions must be auditable, explainable and replicable; but these standards are already a best practice for journalism. 

Provisions that would inform users of their interactions with AI are also good practice for journalists. As journalists use AI, they will have to determine how best to inform the public of its use in order to maintain trust. 

Finally, these proposals make clear that newsrooms can be held liable for any AI tools they develop and deploy, meaning media outlets will be responsible for any damage the AI tools they use cause; although, it is not yet clear what qualifies as prosecutable damage under these proposals. As such, journalists and newsrooms will have to ensure that the AI technology they use follows the above-mentioned guidelines of explainability, auditability and replicability, and newsrooms will have to have policy in place and staff to answer any questions users may have about the newsroom’s AI technology’s decisions. 

Data protection and privacy

AI summary: Countries in the Middle East and North Africa are working to protect data and privacy in their AI policies by updating current laws, creating new ones, or setting up government agencies to oversee data protection. These rules can help protect journalists and their sources, but they may also be used to limit reporting.

Approaches

1. Updating and applying existing legislation. In some cases, such as in Algeria, Egypt and Israel, the AI proposals recommend updating the country’s existing data protection laws to account for generative AI. Other proposals, such as Saudi Arabia’s Generative AI Guidelines for Government and the Public, Qatar’s Guidelines for Secure Adoption and Usage of Artificial Intelligence and Oman’s Public Policy for the Safe and Ethical Use of Artificial Intelligence Systems, outline ways existing data protection legislation can be applied to AI models.
2. Developing a data protection law. According to Libya’s National AI Policy, any processing of personal data must comply with the Personal Data Protection Law as soon as one is enacted.
3. Creating or expanding an oversight authority. Egypt’s National AI Strategy calls for the creation of a personal data protection authority, while Algeria’s National Strategy for AI proposes expanding the Personal Data Protection Agency’s role in overseeing data protection and enforcing AI regulations. 
4. Incorporating privacy into the design of AI models. Saudi Arabia’s Generative AI Guidelines for the Public note that privacy and security should be implemented into AI tools “by design.” Kuwait’s National AI Strategy notes the need to establish a “security baseline” in the development and deployment of AI that would “provide a structured approach to identify vulnerability, establish protective measures and enforce compliance with regulatory requirements.” Similarly, Oman’s Public Policy for the Safe and Ethical Use of Artificial Intelligence Systems requires developers to implement protective measures that ensure data privacy and security and the UAE’s AI Ethics Guide states that AI systems should “respect privacy and use the minimum intrusion necessary.”
5. Providing consent provisions. Saudi Arabia’s Generative AI Guidelines for Public require generative AI service providers to provide users the option to either give or refuse consent to have their data used for AI model training purposes. The guidelines also require service providers to offer an option to remove all prompt history on request, in accordance with relevant laws and regulations.
6. Prohibiting surveillance. The UAE’s AI Ethics Guide states, “Surveillance or other AI-driven technologies should not be deployed to the extent of violating internationally and/or UAE’s accepted standards of privacy and human dignity and people rights.”

Impacts on journalism

As explained in the Sub-Saharan Africa chapter, the African Union’s Continental AI Strategy can provide insight into some ways North African countries are thinking about data protection and privacy. The Continental AI Strategy calls on data and computing platforms to develop policies that facilitate sharing of non-personal data, and it recommends regional governments establish frameworks and protocols that align with the African Union’s Data Policy Framework. It also cites the African Union’s Convention on Cybersecurity and Personal Data Protection, or the Malabo Convention, as an important guiding document. No countries in North Africa have ratified the Malabo Convention, however, meaning that it is not enforceable, but it may still hold normative power in shaping how these countries update their own data protection laws. 

More concretely, data protection and privacy should be built into any AI system newsrooms develop or deploy; this is an important way to protect both end users and journalistic sources, especially in repressive environments. It is worth noting, however, that data protection laws can and have been used to stifle public interest reporting. 

Proposals that prohibit the use of AI for surveillance could theoretically improve the safety of journalists and their sources. In practice, however, the UAE’s laws include exceptions for national security and public interest reasons, leaving the door open for continued government surveillance of journalists. 

Public information and awareness 

AI summary: Countries in the Middle East and North Africa are creating plans to increase public knowledge about AI through awareness campaigns, education, and accessible resources. While these efforts could benefit journalists by giving them the knowledge to cover AI, there are concerns that some governments may try to control how the media reports on the technology.

Approaches

The Gulf Cooperation Council’s Guiding Manual on the Ethics of Artificial Intelligence Use sets the stage for the region by including a recommendation that member states raise awareness of and disseminate knowledge about AI by creating educational materials, running public awareness campaigns and carrying out joint training sessions. Approaches throughout the Middle East and North Africa follow suit. 

1. Running AI awareness campaigns. Algeria’s National AI Action Plan, for example, notes the need to “establish and run an AI Awareness and Literacy Program for the public and decision makers alike, through the media, conferences, workshops, and collaboration with educational organisations.” Egypt’s National AI Strategy calls on the government to partner with the media and influencers “to share positive news about AI,” as well as to partner with media organizations to distribute a survey about public attitudes towards AI. Kuwait’s National AI Strategy calls for public awareness campaigns to “demystify AI, raise awareness about its potential benefits and implications, and empower individuals to make informed decisions in the AI-driven world.” Libya’s National Artificial Intelligence Policy highlights the need for awareness-raising efforts, including through the issuance of press releases on AI news. 

2. Promoting AI literacy. Kuwait’s National AI Strategy features an initiative to promote digital literacy in the country, including among citizens, students and non-technical professionals. Libya’s National Artificial Intelligence Policy notes that AI awareness programs should be developed and implemented for public sector employees. It also features provisions to educate students on AI and to encourage community-wide AI literacy that can help develop smart solutions for the public good. 

3. Providing accessible resources. For example, Kuwait’s National AI Strategy includes an initiative to provide “accessible and user-friendly resources in Kuwait, such as online courses, tutorials, and interactive tools, to facilitate self-paced learning and skill development in AI-related topics.”

Impacts on journalism

AI awareness, literacy and education campaigns could benefit both the public and journalists. An AI-literate public would be better equipped to understand both benefits and harms of AI. These same campaigns could also potentially benefit journalists if they are included; in order to cover AI technology, journalists will need to be up-to-date on the latest developments in the technology industry and to understand how AI companies and their tools are impacting society. Regardless of whether or not it is stated explicitly in the proposals, journalists have an important role to play in sharing information about AI with the public, and it is important they have the knowledge to do so accurately. 

Algeria, Libya and Egypt’s proposals all notably acknowledge the important role of journalists as purveyors of public information; however, Egypt’s call for the media to “share positive news about AI” suggests dictating how journalists cover technology. It is important that policymakers respect the editorial independence of the press, even when partnering with them on public awareness campaigns. 

Conclusion

While journalists and journalism are not central to any AI policies, strategies or laws in the Middle East and North Africa at this time, the existing ones will impact the field. AI offers unique opportunities and risks to journalists in this region. On the one hand, newsrooms can develop and deploy rights-respecting AI to aid in their reporting. This can help newsrooms develop new forms of income, allow reporters to put out content that masks a source’s identity to keep them safe, and help outlets reach new audiences. On the other hand, AI can exacerbate existing threats that journalists face in the region, such as surveillance and technology-facilitated gender-based violence, and vague AI laws can be misused to attempt to silence reporters. 

The post Middle East and North Africa appeared first on Center for News, Technology & Innovation.

Countries in Sub-Saharan Africa view AI technology as a way to develop their economies and contribute to the United Nations’ sustainable development goals, but they also tend to acknowledge that the benefits must be weighed against the risks AI can pose. Many countries across the region also recognize that low internet penetration, low digital literacy and limited access to computing resources can make effective AI implementation difficult. As a result, most of the reviewed proposals in the region take both an innovation-based and a harm-based approach to AI regulation. These strategies encourage their countries to become regional leaders in AI while emphasizing the need to keep humans at the center of this approach. In some circumstances, such as in Kenya, Senegal and Ethiopia, these AI proposals exist alongside other digital development strategies that aim to increase the country’s digital infrastructure and economy. Digital development strategies are not the focus of this analysis, but they are contributing to the development of the ecosystem in which AI and journalism function. 

The African Union’s 2024 Continental AI Strategy is a strong representation of how the region is approaching AI and what it will mean for journalism. This strategy emphasizes AI’s impact on information integrity and media, and it encourages member states to cooperate with each other and take action to mitigate risks in these fields. This strategy was endorsed by the African Union Executive Council, but it is non-binding at the country level, leaving implementation and enforcement to national and regional regulatory bodies. Several countries’ AI strategies and policies are aligned with the African Union’s strategy, such as those of Ethiopia and Lesotho. Other countries, including Nigeria and Zambia, also take inspiration from UNESCO, the OECD and the U.S. National Institute of Standards and Technology Framework for AI Risk Management. 

Between January 2022 and June 2025, 14 countries in Sub-Saharan Africa had adopted AI strategies or policies. Our review did not uncover any AI legislation adopted during this time period, but Namibia is reportedly drafting an AI bill. 

By the numbers

Of the 17 proposals we reviewed in sub-Saharan Africa, three specifically mentioned journalism; none addressed freedom of speech or expression; seven addressed manipulated or synthetic content; 11 addressed algorithmic discrimination and bias; three addressed intellectual property and copyright; 11 addressed transparency and bias; 13 addressed data protection and privacy; and eight addressed public information and awareness. 

Impacts on journalism and a vibrant digital information ecosystem

Each of these seven policy areas has the potential to impact journalism in a way that reverberates across the continent given the cross-border nature of AI. Proposals on public information and awareness could potentially increase journalists’ own knowledge of AI, if the proposed skills trainings include members of the press. AI-generated disinformation campaigns are having a negative impact on the information space and undermining trust in traditional news organizations; legal efforts to counter AI-manipulated content can help protect journalists, but they have also been misused to target journalists in the region.

Proposals that address algorithmic bias and discrimination tend to go hand-in-hand with the journalistic principle of objectivity, but they can also pose a challenge to newsrooms’ efforts to use AI to personalize content. Intellectual property and copyright laws across the region vary widely, and few of the reviewed proposals mention this topic at all. This poses a difficult challenge for news organizations who are losing traffic to AI tools and whose content is being scraped without remuneration.

Transparency and accountability requirements would generally positively impact reporters by providing them with information that would allow them to better understand, and thus better report on, the AI systems that are impacting their local communities. Data protection and privacy regulations are generally best practice for newsrooms, but, when misused, these regulations could be used to stifle public interest reporting. Finally, the omission of freedom of speech and expression in any of the reviewed policies is indicative of declining press freedom in many sub-Saharan African countries.

Freedom of speech or expression

AI Summary: None of the reviewed AI policies in Sub-Saharan Africa mention freedom of speech or expression. Since freedom of speech is linked to a free press, the absence of this topic, along with low freedom of expression ratings in the region, signal a negative environment for journalism.

Approaches

None of the reviewed policies specifically mention freedom of speech or expression. 

Impacts on journalism

The lack of inclusion of freedom of speech and expression in the reviewed AI policies does not necessarily signify that countries in Africa do or do not support this right. Member states of the African Union could embrace the African Commission’s 2019 Declaration of Principles on Freedom of Expression and Access to Information in Africa; the principles include non-interference with freedom of opinion and the protection of journalists. The Declaration is not enforceable, but it does establish a normative framework that member states of the African Union should seek to uphold. 

It is worth noting, however, that, according to Article 19’s 2025 Global Expression Report, no countries in Sub-Saharan Africa rank as “open,” and nearly a quarter of the region’s population lives in a country “in crisis.” Furthermore, attempts to regulate social media platforms in several countries in Sub-Saharan Africa, while often well intentioned, have sometimes resulted in internet blackouts and platform suspensions, which can have negative impacts on freedom of speech, as well as on journalists’ ability to share the news and audiences’ ability to access it. Because freedom of speech and expression creates the environment in which freedom of press can exist, the lack of consideration for freedom of speech paired with the low rankings across the continent can signify a negative environment for press conditions in the AI era. In fact, Reporters Without Borders notes that “press freedom is experiencing a worrying decline in many African nations.” 

Manipulated or synthetic content

AI Summary: While the African Union recommends that countries address AI-driven disinformation with education and new laws, Sub-Saharan African nations have so far only taken a few, often vague, steps to do so. AI creates challenges for journalists, who are needed to check facts but must also deal with a public that has limited trust, while new laws meant to address disinformation could be used against them.

Approaches

One of the African Union’s Continental AI Strategy’s areas of focus is information integrity, media literacy and information literacy. The strategy explains that AI is magnifying mis- and disinformation, and, as such, it recommends that African countries implement media and information literacy programs in schools, train government officials on these skills, develop legal frameworks to regulate emerging technologies and develop strategies to address the risks posed by AI, such as disinformation and hate speech. So far, countries in Sub-Saharan Africa have only lightly touched on these recommendations in their proposals. 

1. Implementing media literacy programs. Kenya’s National AI Strategy is the only reviewed proposal that follows up on the African Union’s recommendations by including an objective to “launch a public awareness campaign on AI rights, disinformation, misinformation, protection and safe development while showcasing the benefits of AI.” 
2. Using AI to detect mis- and disinformation. Both Ethiopia’s National AI Policy and Mauritania’s National AI Strategy 2025-2029 note that AI can be used to help detect and counter foreign propaganda and disinformation campaigns. Neither details how this would be achieved. 
3. Acknowledging the problem without further context. Ghana’s National AI Strategy and Nigeria’s draft National AI Strategy explain that AI-generated content can be used to spread disinformation and manipulate citizens. 

Impacts on journalism

Journalists play an important role in public education about the risks of AI, as well as in fact-checking AI-generated claims and content, such as deepfakes. However, as explained in Kenya’s National AI Strategy, the rise of AI-enabled false narratives can further undermine trust in media and institutions, meaning there might not be a trusting audience receptive to reporters’ fact-checking efforts. 

Legal frameworks can help regulate the use of AI to avoid the spread of misleading, manipulated content, but these frameworks are still evolving across the region, leaving gaps in mitigating these challenges. However, because we have seen “fake news laws” misused to target journalists in the region, such as in Ethiopia, it is important that any AI-related legislation attempting to address disinformation also include provisions to protect journalists and their ability to report. 

Algorithmic discrimination and bias

AI summary: Sub-Saharan African countries are prioritizing AI strategies that focus on addressing bias and making AI systems more inclusive by pushing for the development of AI models in local languages and by prioritizing the representation of diverse populations. To avoid undermining public trust, journalists must be aware of biases in the AI systems they use, while also navigating new regulations that may complicate AI use for personalizing content.

Approaches

AI strategies in the region pay significant attention to algorithmic discrimination and bias. This emphasis is unsurprising given that the majority of existing AI models are trained primarily on English-language data from Global North contexts, while an estimated 1,500 to 3,000 languages are spoken across the region, meaning current, popular AI models may not match users’ experiences or meet their needs in the region. 

The African Union’s Continental Strategy sets the stage for regional attempts to address algorithmic discrimination and bias with its focus on building safe, inclusive AI systems by supporting the development of AI in local languages and developing and implementing AI that is inclusive and beneficial, with an emphasis on reaching women and girls and vulnerable populations. Proposals across Sub-Saharan Africa follow suit and generally advocate for three approaches.

1. Developing AI in local languages and with an emphasis on local, cultural contexts. For example, Côte d’Ivoire’s National Strategy on AI calls for the development of an AI large language model in local languages to better promote and preserve local traditions and knowledge. Kenya’s National AI Strategy is guided by a principle of cultural preservation and contextualization, which states that “AI systems will be developed that are enriched with Kenyan cultural values and that preserve and promote the nation’s cultural heritage and ensure contextual relevance to local needs and contexts.” Kenya’s strategy further explains that to accomplish this, the country is encouraging research aimed at developing AI systems that can interact in various local languages, which, according to the strategy, would democratize access to AI, make AI more relevant to the Kenyan population and contribute to the preservation of linguistic diversity. Similarly, Nigeria’s National AI Strategy includes an objective on driving locally-led AI innovation to “replicate Nigeria’s social context and cultural diversity with AI tools and solutions” to drive accessibility across sectors.

2. Prioritizing inclusion at all stages of the AI lifecycle to ensure AI benefits underrepresented and vulnerable populations. Côte d’Ivoire’s National Strategy on AI includes a strategic principle on societal inclusion and equity, which lays out the need to ensure AI is accessible across geographic and economic boundaries and to all people — especially women, youth and vulnerable groups. Nigeria’s National AI Strategy notes that it is essential that AI innovation is accessible and leaves no one behind. It continues to encourage Nigerians to “actively promote diversity and representation in AI research, development, and deployment” and ensure “that the benefits of AI innovation are shared equitably among all members of society, including marginalised and vulnerable populations.” Ghana’s National AI Strategy notes the need to improve internet and digital infrastructure, particularly in rural areas, to create an environment for inclusive AI in the country.

3. Calling for algorithmic audits and other tests to identify and correct bias in AI systems. Lesotho’s draft Artificial Intelligence Policy and Implementation Plan cites the need to “develop initiatives to identify and reduce biases in AI models, particularly to prevent discrimination and promote equity,” including by mandating regular bias audits that would presumably be reported to the overseeing government entity. South Africa’s National AI Policy Framework also calls for the development of methods to identify and mitigate bias in AI systems, but it does not specify what these methods would be. Côte d’Ivoire’s National Strategy on AI calls for the imposition of algorithmic audits on AI systems to ensure they are not biased; although, how these audits would be reviewed or by whom is not specified. It also encourages diversity in the datasets used to train AI to prevent bias. 

Impacts on journalism

News organizations can play an important role in the development of localized AI models if they so choose. Because media outlets across the region often operate in local languages — sometimes in more than one local language — they have considerable amounts of video, audio and text content that can be used to train localized AI models. However, if news content is used to train AI models, it should be with the original copyright holder’s consent, and forms of remuneration and citation should be considered. 

Journalists and newsrooms throughout the region will have to be cognizant of the biases in the AI systems they develop and use — both for legal reasons and to avoid exacerbating or creating societal tensions. This also aligns with journalism standards since bias undermines the journalistic principle of objectivity. If journalists ignore these biases, they risk undermining public trust in both their reporting and in the technology they use. 

On the other hand, national regulations can pose a challenge to newsrooms’ efforts to use more sensitive data in AI to personalize content if the laws do not include an exception for journalistic uses. News outlets will have to walk a fine line between personalization and bias mitigation. To explain how they do this, newsrooms could consider publishing their own AI ethics policies, which should outline how they are using AI technology, mitigating biases and discrimination, and implementing human oversight of all AI-produced content. 

Intellectual property and copyright

AI summary: Ghana and Nigeria are the only two countries in Sub-Saharan Africa that have so far focused on using existing intellectual property laws to protect the work of AI developers and promote innovation. The lack of clear copyright laws across the region makes it difficult to enforce rules for AI-generated content and makes news organizations’ remuneration efforts more complicated. 

Approaches

While the African Union’s Continental AI Strategy notes that intellectual property regimes are essential to nurturing AI start-ups, so far only two countries in Sub-Saharan Africa emphasize intellectual property and copyright in their own AI strategies. Both countries take a similar approach.

1. Clarifying how existing laws apply to AI. Ghana’s National AI Strategy recommends that the government “review and clarify laws for copyright, patents and intellectual property.” Nigeria’s draft National AI Strategy notes that applying the country’s existing intellectual property laws, including the Copyright Act and the Trademarks Act, to AI is “critical to promoting innovation and protecting the intellectual property rights of developers.”

Impacts on journalism

Copyright laws vary widely across the region, with some laws still focused on the analog era, while others have been updated for the digital era. Furthermore, many countries across the region face difficulties with piracy and the enforcement of intellectual property laws. The cross-border nature of AI poses a difficult challenge for the region; even if one country comes to a copyright decision about a piece of content created with AI or the use of content to train AI, another country might disagree. As countries continue to develop and update intellectual property laws, they should also consider responsibility: When a journalist, or anyone else, uses AI to create content, who owns the copyright of that content, and who is responsible for its potential impacts?

News organizations in several countries around the world, including the United States, Canada and Japan, have sued AI companies over the unauthorized use of their content. News organizations in the United States, Germany and France, among others, have also signed deals with AI companies, allowing the companies to use news content to respond to user queries (with a link back to the news site). As of yet, no news organizations in sub-Saharan Africa have sued or signed deals with AI companies, although South Africa has proposed mandatory content remuneration rules. Experts posit that AI companies are not approaching news outlets in Sub-Saharan Africa for many reasons, one of which is the lack of clear copyright regimes. Stronger regimes, they say, could provide protection for media houses and force AI companies to the table. Furthermore, news outlets that operate in local languages could provide significant swathes of training content for companies building localized AI models — although, these local companies should still seek permission and establish deals with news companies. These kinds of remuneration deals could be an important financial lifeline for journalists in the region whose work is being scraped by AI tools that are not under a legal obligation to link back to the original news source, thus diverting traffic from the original outlet. 

Transparency and accountability

AI Summary: In their national strategies, several Sub-Saharan African countries are pushing for transparency and accountability in AI by creating clear rules for its use and holding developers responsible for how their AI systems work. The new rules would help journalists better report on AI, but they would also require newsrooms to be more transparent about how they use AI and would hold them responsible for the results.

Approaches

Several strategies in Sub-Saharan Africa are guided by the ideals of transparency and accountability, noting that they are essential for building public trust in AI. Country strategies recommend increasing transparency and accountability in three main ways.

1. Establishing clear national standards and governance principles for AI. Benin’s National Artificial Intelligence and Big Data Strategy recommends updating the country’s Code of Digital Affairs to “formalize and institute impact analyses and monitoring of AI solutions throughout their lifecycle.” Rwanda’s National AI Policy states, “​​Trust is critical to public confidence and acceptance of AI. By strengthening the capacity of regulatory authorities to understand and regulate AI aligned with emerging global standards and best practices, we will build transparency and trust with the public.”
2. Requiring developers to explain how AI systems arrive at their outputs. Both South Africa’s National AI Policy Framework and Nigeria’s National AI Strategy prioritize clarity in the design, development, deployment and decision-making of AI systems. This means developers have to explain how the systems work and what kinds of purposes and biases they hold. These principles would also allow developers and deployers to be held responsible for AI’s ethical use. Cote d’Ivoire’s National AI Strategy goes a step further by providing users with a “right to recourse” when they do not agree with an AI model’s decision. 
3. Requiring AI’s decisions to be replicable and auditable. One way to do this, according to Nigeria’s National AI Strategy, is to create an open data initiative to foster collaboration between public and private sectors.

Impacts on journalism

Transparency and accountability in AI systems would generally positively impact the field of journalism. Transparency requirements that necessitate explainable AI would allow journalists to better understand, and thus better report on, the AI systems that are impacting their local communities. By pushing for explainable AI, these same provisions would also benefit the general information ecosystem by increasing the trustworthiness of AI responses; explainable AI prioritizes prediction accuracy, traceability and decision understanding after the results are computed, meaning that users will better understand how an AI system arrived at a result. 

Newsrooms would also be responsible for meeting these transparency standards. As newsrooms develop and deploy AI, they would be expected to be able to explain exactly how the technology works and when, how and why they are using it. They would also be required to provide human oversight over content produced by AI. Depending on the governance principles in each country, newsrooms may reasonably be held liable for any decisions made by the AI technologies they deploy. 

Data protection and privacy

AI summary: Some Sub-Saharan African countries are addressing data privacy and protection for AI systems by either using current data laws or creating new ones to fit AI’s unique features. Newsrooms need to follow these data privacy rules to build public trust, but policymakers should make sure the laws do not prevent journalists from carrying out public interest reporting.

Approaches

The African Union’s Continental AI Strategy calls on data and computing platforms to develop policies that facilitate sharing of non-personal data, and it recommends regional governments establish frameworks and protocols that align with the African Union’s Data Policy Framework. The Continental AI Strategy also cites the African Union’s Convention on Cybersecurity and Personal Data Protection, or the Malabo Convention, which was adopted in 2014 and entered into effect in 2023, as an important guiding document. A handful of AI strategies in the region acknowledge the importance of data protection and privacy in related, albeit slightly different, ways.

1. Applying existing legislation to AI. Nigeria’s National AI Strategy highlights the country’s 2023 Data Protection Act (DPA). The strategy notes that even though the DPA does not specifically address AI, it is still applicable to AI data concepts. For example, AI systems must adhere to the DPA’s principles for data minimization and purpose limitation, and they must protect sensitive data and comply with transparency requirements. 

2. Creating new legal frameworks for AI. Lesotho’s draft Artificial Intelligence Policy and Implementation Plan recommends that the government adopt a comprehensive data protection law, which, the plan states, could align with the EU’s GDPR. Kenya’s National AI Strategy also has an objective to create a robust data governance framework, which would include a specific data policy and establish an AI taskforce within the proposed Data Governance Office Coordination Committee. 

3. Combining the two approaches. Zambia’s National AI Strategy 2024-2026 explains that the country has already established the legal framework to safeguard personal data through its data privacy law. It further acknowledges that Zambia has established the office of the Data Commissioner to ensure compliance with the law and corresponding regulations. The strategy then notes that this legal framework could be evolved to “be fit for purpose” to address AI standards. Mauritania’s National AI Strategy calls on the government to implement policies to comply with data protection regulations. Similarly, South Africa, Benin and Ghana highlight their existing governance structures but recognize that these laws should be strengthened to account for AI.

Impacts on journalism

The Malabo Convention is legally binding in the 16 African Union member states that have ratified it; thus, it provides meaningful insight into how countries across sub-Saharan Africa and North Africa are seeking to handle data protection across borders. However, because it took so long to come into effect, it does not account for AI, and many countries have chosen to adopt more up-to-date data protection laws. Even these more updated laws do not always account for AI, so we chose to focus our analysis on AI-specific strategies, policies and laws. 

Journalists are already well trained on how to protect the privacy of their sources. As such, complying with new and existing data privacy legislation should, generally, fall within existing newsroom practices. Newsrooms should comply with local data privacy legislation and general best practices to ensure data privacy is built into any AI system they use or deploy. In many cases, newsrooms are choosing to publish their data privacy and AI policies publicly to build transparency and increase public trust. 

AI also poses new challenges to journalist and source privacy that are not currently addressed in the region’s approaches. AI can be used to surveil and target journalists and their sources. AI tools can also be used to identify individuals in anonymized data sets, to uncover anonymous authors or sources through language processing and to create a composite image from the blurred face of a source. As policymakers consider how best to apply existing legislation to AI and craft new legislation, they should consider how to protect the data of journalists and their sources from potential harm enabled by AI. 

As in other regions, data protection laws can be misused to stifle public interest reporting; as such, policymakers should consider the impact these laws can have on the news industry and consider exceptions within the laws to protect journalists’ ability to carry out investigations.

Public information and awareness 

AI Summary: Several African countries are creating plans to increase public understanding of AI through public education campaigns, identifying key people in AI management, and teaching the public and workforce new AI-related skills. These plans can also help journalists by improving their digital skills so they can better understand and report on AI.

Approaches

1. Conducting public education campaigns. Kenya’s National Artificial Intelligence (AI) Strategy 2025-2030 highlights a flagship project to launch a public awareness campaign on “AI rights, disinformation, misinformation, protection and safe development while showcasing the benefits of AI” with the goal of increasing the public’s foundational awareness of AI. South Africa’s National AI Policy Framework similarly calls for public awareness campaigns to “educate the public on AI technologies and their implications.” Zambia’s National AI Strategy 2024-2026 lays out a plan to launch AI literacy campaigns to “promote understanding of AI technologies, benefits, risks, and ethical considerations among the public.” Senegal’s National Strategy for the Development of Artificial Intelligence calls on the government to partner with AI technologists, local authorities and civil society to train the public on ways they can use AI to meet local needs. 
2. Mapping out stakeholders in AI governance. Lesotho’s draft Artificial Intelligence Policy and Implementation Plan lists out the roles of stakeholders in AI governance, including those of the government, regulatory authorities, citizens and the media. The media, it notes, is “crucial in shaping public discourse, promoting transparency, and holding stakeholders accountable.” Under this proposal, the media would be responsible for educating the public about AI and its implications, investigating and reporting on potential misuses or ethical breaches in AI systems, and facilitating dialogue between stakeholders and the public. 
3. Training the workforce on AI-related skills. Côte d’Ivoire’s National Strategy on AI calls on the government to carry out national awareness campaigns to explain the benefits of AI to the public servants and explain how they can use the technology to improve public services. In its plan to raise public awareness of AI, Kenya’s National Artificial Intelligence (AI) Strategy 2025-2030 also highlights the need to educate government employees on “ethical, equitable and inclusive AI.” Finally, Nigeria’s draft National AI Strategy includes an objective to “develop a talent pipeline with the necessary knowledge and skills” to increase AI adoption across the country. This includes a plan to develop and implement AI skills development programs across different sectors. 

Impacts on journalism

Journalists play an important role in educating the public about new technologies and their role in society. Lesotho’s policy is the clearest in the region regarding journalists’ role in public education about AI. While other countries may intend a role for journalists in their public awareness campaigns, they do not specify one — perhaps this is a nod to journalistic independence, or perhaps it signifies a lack of awareness of journalism’s role. 

Training the public and the workforce on AI could increase journalists’ own digital literacy if they are included in these programs. This would enable journalists both to effectively implement AI in their own work and to better understand the technology they are expected to report on. 

Conclusion

As countries throughout Sub-Saharan Africa continue to develop and implement AI policies, strategies and regulations, it is important that policymakers think through their impacts on journalism. AI’s impact on journalism does not just affect journalists; it affects the whole of society by either aiding or hindering journalists’ ability to share news with their communities. 
As journalists ramp up use of AI for their own work, newsrooms must ensure that reporters have a full understanding of how the technology works, including the benefits and harms posed by AI, to ensure their reporting remains factual. Journalists must also consider how best to explain their use of AI to the public in order to maintain trust. Furthermore, it will be up to journalists to investigate and report on governments’ and companies’ use of AI to uncover their impacts on society, thus making journalists’ AI literacy even more essential. Finally, as AI proposals are implemented across the region, journalists can determine how best to leverage the technology to better deliver on their mission and to use it as a source of revenue.

The post Sub-Saharan Africa appeared first on Center for News, Technology & Innovation.

Several countries across East Asia and the Pacific have passed (or are actively considering) legislative proposals for regulating AI. Other countries have presented national strategies and guidelines, while some are relying on existing law. The Association of Southeast Asian Nations (ASEAN) has promoted an AI governance and ethics framework for companies looking to develop and deploy AI models in the region (though this is not legally binding). 

There are several notable examples of country-level approaches. For example, China has put forward Measures for Generative Artificial Intelligence, Article 4 of its Measures for Identifying Synthetic Content, which specifically requires identification of synthetic content, and Provisions on Deep Synthesis, which covers internet services and even prohibits the dissemination of “false news.” Additionally, Japan’s innovation-based Act on Artificial Intelligence-related Technologies focuses on research and development. South Korea is another country in the region that has passed comprehensive AI legislation. South Korea’s Basic Act on Artificial Intelligence adopts some ideas from the EU AI Act, though there are notable differences, especially in the realm of classifying risks — South Korea has one broad framework rather than the EU’s four categories. South Korea’s legislation also considers national competitiveness. 

Neither Australia nor New Zealand have a comprehensive AI bill or law as of June 2025, but each country has voluntary frameworks, such as Australia’s Voluntary AI Safety Standard and New Zealand’s Public Service AI Framework, as well as standalone legislation regarding deepfakes either under consideration (the Deepfake Digital Harm and Exploitation Bill in New Zealand) or passed (the Deepfake Sexual Material amendment to Australia’s Criminal Code). Both countries are relying on existing law to regulate AI at the moment. 

Other countries in the region are also considering legislation, including the Philippines (which is also debating a bill on AI development), Taiwan, Thailand (regarding both business operations and AI innovation) and Vietnam (examining the digital technology industry more broadly). Several others have published national frameworks including Hong Kong, Indonesia, Malaysia and Singapore. And while Fiji, Laos, Mongolia and Myanmar have begun planning for AI regulation, they have not yet published official government documents. Most of the smaller island nations in this region do not have existing AI laws or AI frameworks.

Overall, the approaches in East Asia and the Pacific range from innovation-centered to harm-prevention. Few explicitly discuss journalism and the information space. Deepfake and synthetic media regulations, transparency, data privacy and mitigating algorithmic bias receive a large amount of attention, as does increasing technological capacity. 

By the numbers

Of the 23 proposals we reviewed in East Asia and the Pacific, two specifically mentioned journalism; three addressed freedom of speech or expression; 13 addressed manipulated or synthetic content; 13 addressed algorithmic discrimination and bias; nine addressed intellectual property and copyright; 20 addressed transparency and accountability; 17 addressed data protection and privacy; and 11 addressed public information and awareness.

Impacts on journalism and a vibrant digital information ecosystem

The proposals we examined across the East Asia and Pacific region have implications for journalists in each of the seven topics. Public awareness campaigns and educational programs produce opportunities for journalists to not only become more knowledgeable about AI technologies, but also to share information with the public. Many of the proposals, including those in China, South Korea and Vietnam, also explicitly address manipulated content and the transparency measures needed when disclosing to the public that AI has been used to create content. Journalists are already experienced at disclosure given their profession (e.g., disclosing when measures are taken to protect a source’s identity), but with the range of languages and cultures in the region, finding the most appropriate and understandable types of labels may present a challenge to news organizations. 

There are frameworks and legislation in this region considering the implications of algorithmic bias and discrimination. Bias audits can serve as insightful resources for journalists to evaluate training data and algorithmic performance, but these proposals will also require news organizations to update their existing processes to align with legally-binding regulations. 

Data privacy also receives attention in this region’s AI proposals, which will further require news organizations to align their use of AI systems with these regulations. Data privacy requirements are important for protecting personal information, but they may make it more difficult for journalists to investigate training data and for newsrooms to use AI to tailor content to subscribers.

Intellectual property and copyright is integral to news organizations, but the current patchwork of approaches in the region means that international organizations will need to pay special attention to where and when copyrighted training data are legal and where they are prohibited. This variability may also cause journalists in countries with comparatively lax laws on the topic (e.g., Japan) to experience potential advantages when developing novel journalism-specific AI tools compared to their colleagues in more restrictive countries (e.g., South Korea), though it is too early to gather firm evidence on these ideas and is an area for further research.

The topic of freedom of speech and expression is the one area in which the proposals we reviewed contained little information. Given that many countries in the region have low press freedom scores, journalists may encounter challenges reporting on AI if governments or other actors limit the amount of information available to journalists. It is also important to recognize that many of the proposals we examined in the region are non-binding frameworks. Still, future determinations about who is responsible for AI outputs, how legally-binding provisions are carried out, and what penalties exist will have wide-ranging impacts on news organizations and journalists.

Freedom of speech and expression

AI summary: Proposals in East Asia and the Pacific mention freedom of speech but often lack clear rules. Australia’s standard discusses how AI can infringe on civil liberties. The lack of clear rules in AI laws, especially in regions with lower press freedom, could make it harder for journalists to report on and use AI.

Approaches

Several countries’ approaches, including Malaysia and South Korea, mention freedom of speech or expression but lack specific provisions. The Philippines’s draft legislation about deepfakes asserts that the “State recognizes the vital role of free speech in society” but also acknowledges that freedom of expression is not absolute and “carries responsibilities.” Further explanation is not provided. Australia’s Voluntary AI Safety Standard does not call out “freedom of expression” specifically, but it has a section that states “infringements on personal civil liberties, rights …” through the use of AI systems constitute harm to individuals. 

Overall, based on the set of proposals we reviewed, we did not observe freedom of speech and expression outlined in detail; therefore, we did not find concrete categories for approaches in this region. 

Impacts on journalism

The policies about freedoms in East Asia and the Pacific impact journalism in numerous ways. The ambiguous language in this region lacks concrete safeguards for freedom of speech and expression — cornerstones of journalism. These rights may be addressed in other legislation and documents, but their relative absence in AI legislation suggests these ideas are unimportant, especially as most countries in the region (outside of Australia, New Zealand and Taiwan) have lower press freedom scores. Journalists and news organizations may experience greater challenges both reporting on AI and using AI in settings in which press freedoms and press independence are not strongly protected. 

Manipulated or synthetic content

AI summary: Proposals to address manipulated content in East Asia and the Pacific generally prohibit certain types of AI-generated content, can require reports on false information and may mandate labels on AI-generated content to ensure transparency. These policies protect journalists from deepfakes and help them identify AI-generated content, but newsrooms will need to develop appropriate labels and disclosures to follow these laws.

Approaches

Proposals across the region consider manipulated content. They generally fit into three categories. 

1. Prohibiting certain types of AI-manipulated content. Several policies in East Asia and the Pacific prohibit producing (e.g., New Zealand) and/or distributing (e.g., Australia, New Zealand) specific types of manipulated content, such as non-consensual intimate or sexual deepfakes, or engaging in AI-enabled fraud (e.g., Vietnam). These types of proposals typically target the end users of AI tools, not the developers of tools that include this functionality. 

2. Requiring reports on “false information.” China’s deep synthesis provisions require service providers — in this case, generative AI companies that provide tools to create or edit text, video, audio or images — to “employ measures to dispel the rumors” and report the incident to the relevant government department if and when their AI technology is used to create or transmit “false information.”

3. Requiring labels on AI-generated content. There are also important provisions concerning content provenance (i.e., ensuring the authenticity of content) in China, Singapore and South Korea. These countries’ approaches focus on transparency and address labeling AI-generated content. South Korea’s legislation states, for example, “AI business operators shall clearly notify or indicate to users or indicate clearly when virtual sounds, images, or videos are AI-generated, and may be difficult to distinguish from authentic ones.” Arbitration of these decisions will fall under the purview of the Minister of Science and Information and Communications Technology, and methods for disclosure, as well as exceptions, will be prescribed by Presidential Decree. Similarly, China’s regulation requires information service providers to explicitly and “implicitly” (i.e., using techniques users cannot directly observe) identify synthetic media. While China and South Korea passed legally-binding regulations, the government framework in Singapore is not legally binding. Singapore also places the responsibility on service providers and not necessarily end users.

Impacts on journalism

These policies impact journalism in a number of ways. Journalists themselves would be protected from being the subject of non-consensual intimate deepfakes. Because authentication is already central to the journalism profession, requirements to label and disclose AI generated content could help journalists more quickly identify this type of content. 

In the same vein, mandated labeling and transparency provisions will also apply to newsrooms across a range of content including text, images, audio and video. Journalists already regularly disclose altering voices and faces to protect identities, so these requirements are unlikely to pose a major burden on most news organizations in these cases, though this use case is not globally viewed as best practice. That being said, appropriate labels, metadata tags and other forms of provenance/disclosure will need to be developed to comply with relevant laws. The legislative examples from the region include guidelines for how these disclosures should be made, but how audiences will interpret and understand these disclosures needs further consideration. 

Algorithmic discrimination and bias

AI summary: Proposals on algorithmic discrimination and bias in East Asia and the Pacific offer general guidance, establish bias audits, prohibit discriminatory AI uses and plan for government agencies to create anti-discrimination policies. These approaches carry the potential to increase scrutiny on how newsrooms use AI for content recommendations and personalization, requiring them to develop bias prevention plans. They may also provide journalists — as well as everyone, in general— more opportunities to examine AI systems for bias.

Approaches

Many of the proposals in this region discuss algorithmic discrimination and bias. They generally fit into four broad categories. 

1. Offering guidance and information about safe uses. Australia’s Voluntary AI Safety Standard highlights the need for guardrails to protect against AI harms in various realms (e.g., employment and healthcare) by focusing on inclusion and fairness. Relatedly, Australia’s Guidance on Privacy and Developing and Training Generative AI Models includes information about how AI systems can produce biased and discriminatory results based on their training data. We observed similar perspectives in the Hong Kong, Malaysia and ASEAN frameworks. 

2. Establishing bias audits. Proposals in both Malaysia and Singapore suggest regular audits of developers’ AI systems to support responsible AI use and to mitigate AI biases and discrimination. 

3. Prohibiting the development and deployment of AI technologies for discriminatory purposes. China’s Measures for Generative Artificial Intelligence Services regulation requires that “[d]uring processes such as algorithm design, the selection of training data, model generation and optimization, and the provision of services, effective measures are to be employed to prevent the creation of discrimination such as by race, ethnicity, faith, nationality, region, sex, age, profession, or health.” An explicit outline of these measures is not provided in the regulation. Similarly, a foundational principle in Vietnam’s legislation is that AI systems are not to be used to discriminate. 

4. Stating that policies to prevent AI discrimination will be created by government agencies. Section 5b in the Philippines’ draft legislation aims at preventing algorithmic discrimination by including a Bill of Rights, though the official policies to prevent algorithmic discrimination would be specified by the Philippine Council on Artificial Intelligence if and when the legislation passes. Similarly, Taiwan’s draft proposal states that the government, through the Ministry of Digital Affairs, shall prevent algorithmic bias.

Impacts on journalism

These policies impact journalism in several ways. Policies concerning algorithmic discrimination and bias will likely increase scrutiny of how newsrooms use AI to recommend and personalize their audience’s content. For example, a content personalization system might be biased against certain kinds of users by reinforcing stereotypes or only presenting certain kinds of content. Thus, newsrooms may also be tasked with creating AI bias prevention plans for the AI systems they develop and deploy in the region if these plans do not already exist. 

Auditing requirements may be beneficial for journalism since journalists could have an easier time examining AI systems and their potential for bias and discrimination.

Intellectual property and copyright

AI summary: Proposals for intellectual property and copyright related to AI are still being developed and legally decided in the region, with some proposals adhering to existing laws, others allowing copyrighted works to be used for AI training with exceptions and some requiring licenses. The range of approaches and varying stages of policy development creates challenges for news organizations, as there is confusion about who owns AI-generated content, whether they need to pay for copyrighted material to train their AI and how to deal with different laws across various countries.

Approaches

There are several examples of policies and frameworks in the region that reflect considerations for intellectual property and copyright. However, they are generally not detailed, and several proposals state that specific provisions will be developed later. They generally fit into three categories. 

1. Respecting and protecting existing copyright and intellectual property regulations. China’s generative AI regulation stipulates that both deployers and developers need to comply with intellectual property laws in the country. A recent lawsuit found that using copyrighted data to train AI models was legal if there was no intent to plagiarize the original work, while the legal ramifications of using copyrighted works to train AI are currently being reviewed. A separate court case in China found that images generated with AI tools can be copyrighted.

2. Permitting copyrighted works for training AI systems. Japan’s Agency for Cultural Affairs released updated guidance on AI and copyright in early 2024, through which the country generally espouses that copyrighted works can be used for training, unless the copyrighted data are pirated and the AI provider knowingly used pirated data. 

3. Requiring licensing from copyright holders for training AI systems. Taiwan’s Intellectual Property Office has a different perspective, stating that, without prior licensing from the copyright holder, using copyrighted materials for model training violates the country’s copyright laws.

Impacts on journalism

Overall, intellectual property considerations, while present in this region, do not necessarily reflect a common understanding of what constitutes violation of intellectual property rights. Although countries in this region do have preexisting copyright approaches, it often remains unclear who owns the content AI systems generate.

These approaches impact journalism in a variety of ways. For one, ambiguity about the ownership of AI-generated content affects news organizations because their content has been used to train large language models and AI systems. Some contend training AI systems with news content is related to the idea of fair usage. Australia’s government, on the other hand, recently rejected a proposal that would have allowed technology companies to mine creative content to train their AI models, preventing the “large-scale theft of the work of Australian journalists and creatives.” It is also unclear if news organizations hold the copyright to content they may have produced with the aid of AI tools. 

Whether news organizations are allowed to use copyrighted material to train their own models or need to sign licensing deals with copyright holders adds a financial constraint to news organizations. The lack of interoperability in existing copyright laws and approaches across the region may also pose problems for news organizations operating in more than one country as they develop AI models and AI-generated content. News organizations may also need to verify that third-party AI tools developed abroad do not violate their country’s intellectual property laws; however, the policies reviewed for this report address this very little. 

How countries in the region will enforce intellectual property and copyright provisions, which could range from banning AI products in the country to fining or even imprisoning end users, remains to be seen. Ultimately, what data are acceptable for training and who is held responsible for AI outputs will have wide-ranging consequences for news organizations across the region by (1) shaping what data can be used for training AI models, (2) determining if licensing or other forms of compensation are required for using data and (3) setting the penalties for breaking the developing regulations. 

Transparency and accountability

AI summary: Proposals in East Asia and the Pacific focus on transparency in AI use by requiring labels on AI-generated content, providing notice to users, offering non-technical explanations, developing new legal frameworks, ensuring human oversight and monitoring AI systems. These policies can assist journalists by making it easier to identify and explain AI-generated content to their audiences.

Approaches

Most of the approaches in the region discuss the importance of transparency and accountability when using AI, including frameworks in Australia, Hong Kong, Indonesia, Malaysia, New Zealand and Singapore; draft legislation in the Philippines (including draft deepfake legislation), Taiwan, Thailand and Vietnam; and passed regulations in China, Japan and South Korea. They generally fit into six categories. 

1. Requiring labeling and disclosure for AI-generated content. Policies regarding synthetic media and deepfakes, such as China’s regulation on synthetic content and the Philippines’ Deepfake Accountability and Transparency Act, include specific requirements for disclosing AI content. China’s regulation and South Korea’s 2025 AI Act are the two proposals we reviewed that also address generated text. Many news organizations, though, already have internally developed AI disclosure policies.

2. Providing notice to end users about AI systems. Approaches in South Korea and the Philippines suggest or require that users be notified in advance that they will (1) be engaging with or (2) have their data used by an AI system.

3. Including non-technical explanations for the public. Several proposals include provisions to provide access to clear, comprehensive explanations about what AI systems do, how they work and what protections are in place. Australia’s framework focuses on the activities of AI developers, whereas Malaysia and Hong Kong’s frameworks focus more on deployers.
4. Developing new legal frameworks. Malaysia’s framework states that the country should create a legal structure that “assigns responsibility for AI systems” and clarifies how to enforce these new provisions.
5. Holding AI accountable through human oversight. New Zealand’s public service framework stresses the importance of AI being overseen by “accountable humans with appropriate authority and capability.”
6. Requiring AI system monitoring and auditing. A draft approach in the Philippines calls for developers and creators of AI systems to enable monitoring and auditing to “ensure that entities deploying AI technologies are accountable for their consequences,” with specific provisions being developed by the AI Board, which will be created by the legislation. Language about the importance of auditing AI systems also appears in Malaysia’s framework.

Impacts on journalism

Policies covering transparency and accountability impact journalism in several ways. Some of these approaches mandate AI disclosure. This is an opportunity to build audience trust about the AI tools used by journalists, but it also runs the risk of decreasing trust if not done in a way that the public can understand and see the value. News organizations will also need to develop disclosures and labels that use plain language. This will likely look different across the region due to distinct languages, cultures and experiences. News organizations will likely need to revise and add to existing ethics codes and style guides to codify transparency and accountability efforts related to AI models and systems.

Transparency and accountability can benefit journalists’ reporting on emerging technologies by providing greater access to training data, model development and overall AI system performance. Bias audits serve as an important opportunity for journalists to evaluate AI systems and convey how these systems work to the public. Overall, labels and disclosures, and transparency about AI systems will help journalists identify and report on AI-generated content, and explain this content to their audiences. 

Data protection and privacy

AI summary: Proposals in East Asia and the Pacific encourage companies to use personal data responsibly and anonymize data. They also strengthen existing data protection laws, give individuals control over their data and create government investigation protocols for violations. These approaches will require stricter handling of personal data by news organizations (and, in fact, all sectors), potentially leading to challenges with cross-border data security and making it harder for journalists to evaluate AI systems if training data remain inaccessible.

Approaches

There are many examples from East Asia and the Pacific regarding AI and data protection/privacy. These generally fit into five categories. 

1. Urging companies to use personal data fairly and responsibly. Through nonbinding frameworks, Australia and Singapore urge that personal data be used fairly and responsibly by articulating how existing personal data privacy laws apply to AI. New Zealand’s Public Service AI Framework views data privacy as “a core business requirement” for those planning to distribute and use AI systems. 

2. Anonymizing personal data. Vietnam’s recent digital technology law stipulates that “organizations and individuals sharing, exploiting and using digital data in the digital technology industry [which includes AI systems] are responsible for anonymization of digital data unless otherwise prescribed by law.” The legislation includes similar requirements for protecting personal data.

3. Strengthening and supporting existing data protection laws. Malaysia’s framework asserts the opportunity to further develop laws in the country. These protections would involve “setting standards for obtaining informed consent, ensuring data security, and defining the permissible uses of personal information.” The AI Development and Regulation Act proposal in the Philippines stipulates that personal data must be protected in accordance with existing laws.

4. Providing individuals with control over their data. Hong Kong’s framework includes data privacy as one of its 12 ethical principles, with a specific focus on individual rights over personal data. Indonesia’s circular letter takes a similar approach to Hong Kong. The Philippines’s “Right to Privacy” section of its AI Regulation Act has strict requirements for protections of personal data, including that “only data strictly necessary for the specific [AI] context is collected” and that designers, developers and deployers “shall seek permission and respect the decisions of every person regarding collection, use, access, transfer, and deletion” of their personal data.

5. Creating government investigation protocols for violations. Article 16 in Japan’s recent AI law asserts there will be a government investigation when “the rights and interests of citizens have been infringed” as a result of an AI system, though who starts this review and how it is done are not specified in the legislation.

Overall, countries in the region have pre-existing personal privacy and data protection laws; the examples above highlight countries considering how developments in AI intersect with these laws. Many of the countries and territories in this region mention these previous laws in their current AI approaches — particularly those with frameworks — and issue calls to update existing laws to incorporate AI technologies. 

Impacts on journalism

These approaches impact journalism in important ways. The proposals we reviewed will require stricter handling of personal (potentially audience) data when developing and deploying AI models. This requires journalists and newsrooms to develop clear protocols for how to protect and handle sensitive data when using AI tools. There may be challenges for news organizations that operate in more than one country regarding how they handle cross-border data security because not every country in the region will have the same laws and regulations.

Journalists may also experience difficulty evaluating and covering AI systems if training data and algorithms are private due to personal data security practices. Developing ways to examine AI systems, while at the same time protecting personal data, will be important for journalists. 

Public information and awareness 

AI summary: Proposals for public awareness about AI include formal education, public campaigns and promoting AI technologies. These approaches will increase the need for AI literacy among the public and news organizations, requiring journalists to explain AI technologies and potentially collaborate on public education initiatives.

Approaches

There are several examples across the region that speak to the idea of public information and awareness across both policy frameworks and legislation. The proposals we reviewed generally fit into four categories.

1. Developing formal educational programs. Indonesia, Japan and the Philippines are focusing on formal educational programs to build AI knowledge and capacity. These education programs are intended to develop AI knowledge and capacity as well as an understanding of AI ethics. However, exactly who teaches whom about AI and how this education will unfold has not yet been specified. 

2. Creating public awareness campaigns. The Philippines proposes to institute “a nationwide information campaign with the Philippine Information Agency (PIA) that shall inform the public on the responsible development, application, and use of Al systems to enhance awareness among end-consumers.” Proposals from Malaysia and South Korea contain similar approaches.

3. Tasking AI developers and deployers with raising awareness about AI. Hong Kong’s Ethical AI Framework emphasizes that it is important for stakeholders (presumably developers and deployers) to communicate with end-users about what risks exist in AI models and how they will address these risks. The framework also touches on “educating the public to build trust.” 

4. Promoting AI technologies. Japan, South Korea and Taiwan also position the government as a key player in shaping public awareness of AI and include provisions about how the government needs to promote AI technologies across education, industries and the public sector. In a similar vein, Singapore’s Governance Framework for Generative AI includes a provision about “AI for Public Good” that prioritizes access to AI technologies and responsible AI use. 

Impacts on journalism

Journalism is not mentioned specifically as either an audience or vehicle for public awareness campaigns. Yet, these policies can impact journalism in a number of ways. There will be a greater demand and need for AI literacy for both the public and news organizations as AI becomes more common. Journalists will be partly responsible for explaining AI technologies and their uses to the public; thus, developing AI knowledge is critical for news organizations. Public awareness campaigns may include journalists as an audience and may provide journalists with the opportunity to learn about AI and how it can aid their work.

There will be opportunities for collaboration on public education between governments, non-governmental organizations and journalism organizations. However, given disparities in technological access and usage, regional variations will likely mean some journalists will have lower AI knowledge, leaving them less able to effectively assess and explain these technologies to the public; although, technology companies or non-governmental organizations may attempt to fill this gap with their own AI training programs. CNTI’s Global AI in Journalism Research Working Group recently found that there is a dearth of research focusing on East Asia & the Pacific. 

Conclusion

Countries in East Asia and the Pacific are in various stages of AI regulation. China has had regulations on AI, synthetic media and deepfakes for several years. Other countries have recently passed legislation (e.g., Japan, South Korea, Vietnam), are actively considering AI legislation (e.g., the Philippines, Taiwan) or are relying on frameworks and existing laws (e.g., Australia, Indonesia, Malaysia, New Zealand). AI is an ongoing area of focus for countries in East Asia and the Pacific and developing regulations and legislation will continue at a quick pace.

While no policy or framework reviewed for this report explicitly focuses on news and journalism, the approaches will still impact those working in the industry. There are opportunities for news organizations to continue building tools for their own use and to build trust with audiences as proposals in the region address transparency, bias and manipulated content. 

Five of the seven topics we examined in this report — transparency and accountability; data protection and privacy; algorithmic discrimination and bias; manipulated or synthetic content; and public information — are discussed in at least half of the reviewed policies, indicating most proposals in the region are reckoning with these ideas.

Freedom of expression, compared to the other six attributes we looked at, gets significantly less attention, which may put journalism at risk. Other shortcomings in this region include vague enforcement mechanisms and unclear guidance on how to fulfill disclosure requirements. Related to these ideas is the question of who is legally responsible for AI outputs — some of these details will be decided by councils or government bodies created by bills/legislation.

The post East Asia and the Pacific appeared first on Center for News, Technology & Innovation.

One of the most economically dynamic parts of the world, South Asia is home to a young, booming, technically literate population; technology companies willing to invest increasing resources in what they term ”emerging” and “frontier” markets; and governments toeing a fine line between caution and optimism about technology policy, particularly artificial intelligence. Most governments in the region have refrained from rushed regulatory proposals about AI; thus, the consequences for journalism are challenging to predict. 

Recently surpassing China as the world’s most populous country, India has attracted the attention of global technology companies, drawn not just by the potential size of the market but also by a strong higher education sector, a technically skilled workforce, widespread proliferation of low-cost technology and a regulatory environment that has, in the last couple of decades, aimed to open up the country to more foreign direct investment. India’s approach to regulation is deliberate: the country aims to balance risks, innovation, human rights and local economic development. 

Bangladesh, the region’s second largest economy, pushes for both innovation and harm-prevention in its AI approach. Bhutan, Pakistan and Sri Lanka have introduced an AI policy, law, and strategy, respectively.

There are no laws or policies impacting the entire region; although, it is worth keeping an eye on regional blocs for any future convenings and policy developments. In May 2025, Bay of Bengal Initiative for Multi-Sectoral Technical and Economic Cooperation (BIMSTEC), which includes Myanmar and Thailand but excludes Pakistan, Afghanistan and the Maldives, convened a workshop on “AI Standards for Efficiency in Telecommunications & ICTs: Shaping the Future Responsibility” in New Delhi in collaboration with the International Telecommunication Union. Aimed at promoting development of inclusive and ethical AI standards, the workshop was meant to be the first in a series to lay out pathways for future public-private collaboration in “creating technical standards to ensure AI systems are secure, interoperable, and globally aligned.” The South Asian Association for Regional Cooperation (SAARC) has not yet announced or convened any overarching AI guidelines for formal adoption across all member states.

By the numbers

Of the eight AI proposals we reviewed in five countries in the region, one specifically mentioned journalism (or media); two addressed freedom of speech or expression; five addressed manipulated or synthetic content; eight addressed algorithmic discrimination and bias; five addressed intellectual property and copyright; seven addressed transparency and accountability; seven addressed data protection and privacy; and four addressed public information and awareness.

Impacts on journalism and a vibrant digital information ecosystem

Each of these seven policy areas has the potential to have significant impacts on journalism and the information space across the region. Public information campaigns and related training could provide journalists with the skills and knowledge they need to effectively use and report on AI. Provisions encouraging human review and content labelling of AI-generated content could help journalists ensure they do not accidentally spread misinformation. In order to uphold the journalistic standard of objectivity, journalists and newsrooms will have to be cognizant of the biases of the AI systems they develop, deploy and report on. Like other regions, journalists in South Asia are facing difficulties when it comes to copyright in the AI era. The proposals we reviewed could help redirect traffic to news sites by requiring civil servants to double check AI responses and to cite their sources; they would also require journalists to take similar steps when using AI. In general, transparency and accountability requirements for AI would allow journalists to better access and understand AI systems’ data, thus enabling them to better report on the technology, but the proposals we reviewed currently fall short. As countries throughout South Asia further develop their data governance frameworks, newsrooms will be required to comply with these local regulations to ensure data privacy is built into any AI systems they use or deploy. Finally, the lack of consideration for freedom of speech in the reviewed AI proposals paired with the low press freedom rankings across the region could signify a negative environment for press conditions in the AI era. 

Freedom of speech and expression

AI summary: Of the AI proposals reviewed in South Asia, only two mention freedom of speech or expression. Sri Lanka’s strategy mentions the need to balance controlling harmful AI content with protecting freedom of expression, and India’s Fairness Assessment highlights that AI systems should not impact individual rights. The fact that most AI proposals don’t mention freedom of speech, combined with the low press freedom rankings in the region, suggests a negative environment for journalism and upcoming laws in India could make it easier to censor journalistic content.

Approaches

Two AI proposals in the region mention freedom of speech or expression. They do so in different ways.

1. Balancing the creation of harmful content with freedom of expression. Sri Lanka’s National Strategy on AI states, “Generative AI can accelerate the creation and distribution of misleading and harmful content. However, any efforts to regulate this must carefully balance the need to protect the public with the importance of upholding freedom of expression and ensuring access to information.” It does not go into further detail. 
2. Assessing AI’s impact on constitutional rights. India’s Fairness Assessment and Rating of Artificial Intelligence Systems states that AI systems should be classified as high, medium or low risk based, in part, on how “the outcomes affect users’ rights and freedom as per the constitutional and ethical considerations.” While it does not specify freedom of expression, it is implied since it is a guaranteed right under Article 19 of India’s Constitution. If an AI system is found to be high-risk if it is anticipated to impact an individual’s rights or freedom. This assessment is voluntary, however, so it is up to the company to determine whether they adhere to the findings from this standard. 

Impacts on journalism

The omission of freedom of speech and expression in the reviewed AI policies does not necessarily mean that countries in South Asia do or do not support this right — nor does it signify that Sri Lanka and India care more for free speech than their neighbors. However, none of the countries in this region rank “open” in Article 19’s Global Expression report. Sri Lanka was the only country in the regionto improve its freedom of expression ranking in the last year, now coming in at “less restricted,” while India is categorized as “highly restricted.” Because freedom of speech creates the environment in which freedom of press can exist, the lack of consideration for freedom of speech in the reviewed AI proposals paired with the low rankings across the region could signify a negative environment for press conditions. Reporters Without Borders’ 2025 World Press Freedom Index reinforces this, noting that media concentration and political collusion threaten press freedom in several South Asian countries.

Manipulated or synthetic content

AI summary: Government proposals are focusing on human review of AI-generated information, requiring platforms to stop illegal content from being shared and suggesting that AI-created content be labeled. These regulations could help journalists by promoting transparency and accuracy, but some of the rules could also be used to censor them or create confusion about what is real.

Approaches

AI strategies, laws, and policy proposals in the region generally address manipulated or synthetic content in three ways. 

1. Calling for human review. Bhutan’s Guideline for Generative Artificial Intelligence Usage in the Civil Service notes the essential need to address AI-generated misinformation. The guideline explains that while generative AI might answer questions, those answers might not always be accurate; therefore, it is incumbent upon civil servants to verify and review AI-generated content before official use. If misinformation is spread due to a user’s carelessness, that individual faces legal liability for any harms caused. 
2. Requiring platforms to ensure AI does not make it possible for users to share unlawful content. India’s March 2024 AI Advisory ​​requires intermediaries and platforms to ensure that use of AI models, large language models and generative AI “on or through its computer resource” does not make it possible for users to “host, display, upload, modify, publish, transmit, store, update or share” unlawful content.
3. Calling for content labeling. India’s March 2024 AI Advisory further advises that if an intermediary permits or facilitates the creation of synthetic content or the generation or modification of text, audio or visual information, which could then be used as misinformation or a deepfake, they also label that content or embed it with a permanent unique metadata or identifier. 

Impacts on journalism

Bhutan’s requirement of human review is not only applicable for civil servants; this same concept should apply to journalists who use AI for their own work or who report on AI outputs. By reviewing AI-generated content, journalists can ensure its validity, thus ensuring they do not contribute to the spread of misinformation. Likewise, content labeling can increase transparency in the information space and is generally a good practice for news organizations. However, because this portion of India’s AI Advisory is not legally binding, not all organizations will follow the labelling suggestion. This could lead to a split between organizations that choose to adopt these measures and those who do not, which could lead to confusion in the information space and inconsistent disclosure about content that is AI-generated. However, these rules in India’s AI Advisory will become binding if the proposed Draft Rules on Synthetic Labeling, which are currently under public consultation, become public (these draft rules were introduced in October 2025 and, because they fall outside the timeframe covered in this paper, were not analyzed in greater detail).

The provision in India’s AI Advisory requiring platforms to prevent users from sharing unlawful content is voluntary, but it does point back to the Information Technology Act of 2000, which has been used to block the social media accounts of journalists and news organizations. 

Algorithmic discrimination and bias

AI summary: Several proposals in South Asia aim to prevent bias in AI systems by requiring companies to check for fairness during development, providing certification for fair AI and teaching people how to avoid and identify biased content. These efforts will require journalists to be aware of biases in AI tools they use or report on, and equitable access to AI could help journalists by allowing them to use AI tools in local languages.

Approaches

Several AI proposals in South Asia are attempting to address algorithmic discrimination and bias, with varying levels of commitment.

1. Preventing bias by design. Bangladesh’s National Artificial Intelligence Policy states that, by design, AI systems must “prevent from unintended direct/indirect prejudice, bias, and discrimination,” noting that to ensure this, AI companies must carry out compulsory evaluations, consider the diversity of those evaluating AI systems and assess potential risks during the procurement process. Sri Lanka’s National Strategy on AI notes the need to monitor data for potential biases. India’s March 2024 AI Advisory calls on intermediaries and platforms to ensure their AI models and algorithms do not permit bias or discrimination, but it does not explain how they should do this. 
2. Certifying fairness. India’s Fairness Assessment and Rating of Artificial Intelligence Systems provides a systematic approach to certifying fairness for AI systems. It “approaches certification via a three-step process involving bias risk assessment, threshold determination for metrics, and bias testing.” Third-party auditors could provide this fairness certification to AI companies. 
3. Avoiding bias in use. Bhutan’s Guideline for Generative Artificial Intelligence Usage in the Civil Service encourages civil servants to learn about bias, diversity, inclusion, anti-racism, values and ethics to improve their ability to identify biased content. It also notes that civil servants should be careful of the prompts they provide to generative AI systems to prevent the generation of biased or discriminatory content. 
4. Ensuring equitable access to AI. Sri Lanka’s National Strategy on AI further states that the country will develop and deploy AI solutions that “address societal challenges, improve quality of life, and distribute benefits equitably across all segments of society. We will prioritize responsible innovation, aligning AI technologies with ethical standards, human rights, and societal values …” Pakistan’s proposed Regulation of Artificial Intelligence Act similarly states that the national AI commission shall “ensure equitable access and equal opportunities in the field of Artificial Intelligence to all citizens regardless of difference in religious, gender, ethnic, geographic, financial, or physical abilities” but it does not specify how. While not a policy, the Indian government launched both the IndiaAI Mission, which aims to “democratize computer access,” among other things, and BHASHINI, an “open-source multilingual AI initiative supporting 22 Indian languages,” which is a response to the challenge of securing inclusive data that reflects the diversity of Indian society. 

Impacts on journalism

In order to uphold the journalistic standard of objectivity, journalists and newsrooms will have to be cognizant of the biases of the AI systems they develop, deploy and report on. If newsrooms are developing their own AI tools, they should consider bias and other risks at every stage of the AI lifecycle to ensure they are not exacerbating societal divisions. If newsrooms are using AI systems, they can follow the steps described in Bhutan’s guide for civil servants to try to prevent biased outcomes; however, these efforts can only go so far if an AI model is trained on biased data. 

By ensuring equitable access to AI, governments can enable increased AI uptake throughout their societies; journalists, however, will act as important watchdogs as they report on how these technologies are used and impact society. The Indian government’s technology proposals may aid the independent efforts of citizen journalists by allowing them to use AI tools in local languages; industry collaborations with newsrooms could help prioritize algorithmic transparency in emerging AI tools. 

Intellectual property and copyright

AI summary: Governments in South Asia are in the early stages of updating their intellectual property and copyright laws for the AI era, with Indian news outlets’ legal cases against major AI developers potentially providing a crucial test case for the entire region.

Approaches

A few countries in South Asia recognize the difficulties of establishing and enforcing copyright in the AI era.

1. Guiding companies on Intellectual Property protection. In its National Strategy on AI, the Sri Lankan government details how it will provide “technology transfer services” to AI companies, which will explain guidance and support throughout the commercialization process, from intellectual property protection to business model development, to help academics and researchers “spin out their AI innovations into start-ups.”
2. Updating copyright framework. Bangladesh’s National Artificial Intelligence Policy notes that the National Strategy for AI will include an updated intellectual property framework to “incorporat[e] ownership rights, patents, and copyright regulations regarding AI models, AI-generated works, source code, and data.” 
3. Citing sources to avoid copyright infringement. Bhutan’s Guideline for Generative Artificial Intelligence Usage in the Civil Service explains that the sources upon which AI models are trained may be copyrighted and, as such, these models may reproduce copyrighted content in their output. The guideline notes that because generative AI does not always include citations in its outputs, it is up to civil servants to find and cite the right sources to avoid copyright infringement. 

Impacts on journalism

Bhutan’s guideline hits on a significant tension facing journalism in the AI era: news content is often used to train AI models, but these models often do not cite the original source in their output, nor do AI companies always pay the copyright holders for use of the data. This can result in decreased traffic to news sites and revenue loss. On the other side of this tension lies journalists’ own use of AI: like civil servants, if journalists are using AI to gather information, they should confirm its veracity and find the original source to cite. 

Sri Lanka’s plan to work with start-ups on AI could benefit news outlets who attempt to develop their own AI models, and Bangladesh’s plan to update its copyright framework could also provide much-needed clarity for how journalistic content is used in the AI era. 

AI and copyright are still in flux in the region, as demonstrated by the lawsuits between news outlets and AI companies. India-based Asian News International (ANI), NDTV, Network18, the Indian Express, the Hindustan Times and the Digital News Publishers Association (DNPA) have sued OpenAI for allegedly using their news content to train ChatGPT without permission. OpenAI claims that this content falls under the “fair use” doctrine, as per Section 52 of India’s Copyright Act, which would allow the company to use the content without permission or payment to the copyright holder. However, the new India AI Governance Guidelines, which were published after the data collection period for this report, state that “fair dealing” exceptions “may not cover many types of modern AI training” — although the Department for Promotion of Industry and Internal Trade has yet to release its final recommendation. Further complicating the issue, India’s Personal Data Protection Act, which is in the process of being operationalized but falls outside the scope of this paper, includes a fairly expansive view of how companies can collect public data.

Similar court cases are ongoing in the United States, United Kingdom, Japan and other countries. In other jurisdictions, news organizations have addressed this copyright issue by forming paid partnership agreements with AI companies, but so far no such partnerships have been established in India or other countries in South Asia. The outcome of this legal case in India will likely be a bellwether for how publishers, AI companies and lawmakers in the rest of the region attempt to address this issue. 

Transparency and accountability

AI summary: India, Pakistan and Bangladesh are working on AI policies to ensure transparency and accountability by certifying fairness, setting ethical standards and requiring documentation and audits. These efforts could help journalists by providing greater access to information about how AI systems work, which would allow them to better report on the technology and its limitations.

Approaches

AI strategies, laws, and policy proposals in the region generally address transparency and accountability in three ways. 

1. Certifying fairness. India’s Fairness Assessment and Rating of Artificial Intelligence Systems states that governments, businesses and non-profits can pursue fairness certifications to demonstrate their efforts to root out bias in AI systems, which can help build “trust, equity, and transparency among the people.” It further states that “citizens would be the beneficiaries of fairness certification.”

2. Establishing ethical standards. Pakistan’s Regulation of Artificial Intelligence Act, for example, calls on the national AI commission to “make the processes and procedures of collection, storage, and usage of data in Artificial Intelligence systems more accountable, transparent and translućent,” as well as to “ensure implementation on the principles of accountability, privacy, safety, security through usage of Artificial Intelligence technologies.” Likewise, Sri Lanka’s National Strategy on AI states that the government “shall establish robust ethical standards, privacy measures, and security protocols in line with international frameworks to create a trustworthy AI environment. We will ensure transparency in decision-making processes, respect individuals’ privacy rights, and uphold principles of fairness and nondiscrimination.” It does not specify how this would be achieved. 

3. Requiring documentation and audits. Bangladesh’s National Artificial Intelligence Policy requires AI systems, by design, to document the datasets, processes and algorithms “in a standard way to allow for traceability and transparency.” It also requires AI systems to establish “mechanisms to ensure AI compliance with regulations and standards through regular audits and assessments.” The policy does not specify how this would be achieved. 

Impacts on journalism

Transparency is not a panacea for building more trust in journalism, but it can go a long way in helping journalists be more upfront about the strengths and limitations of their sources and their technology-assisted reporting processes. In general, transparency and accountability for AI would allow journalists to better access and understand AI systems’ data, thus enabling them to better report on the technology. 

Pakistan’s AI Regulation Act, however, currently falls short; it mentions algorithmic transparency but does not mandate explainability or accountability for AI-driven decisions. Thus, AI companies are not required to follow these steps, meaning the data might not be accessible to journalists or the public. 

Bangladesh’s policy requiring documentation of algorithmic processes could have implications for journalism chatbots and more public-facing news interface technologies. These implications pertain to data collection and processing, purpose limitation, and data security, as is the case in Europe with the General Data Protection Regulation (GDPR). Future legislation, if building on Bangladesh’s policy, could further strengthen transparency measures by ultimately making individuals aware that they are interacting with AI (e.g., chatbots), or that content they are interacting with was AI-generated. Bangladesh’s policy may also aid the work of journalists more broadly by making it easier for them to report on AI systems and their developers.

Stronger transparency obligations and enforcement could also encourage news organizations themselves to be more transparent about their sources, data and even financials. Research conducted in India, the United States, the United Kingdom and Brazil suggests such transparency measures can make readers more likely to trust news organizations.

Data protection and privacy

AI summary: To ensure the safe and ethical use of AI, countries in South Asia are working to establish data governance frameworks, strengthen data protection, and build privacy and data security into the AI lifecycle. As these regulations are developed, newsrooms will need to follow them to protect their reporters and sources; however, these regulations could be misused to stymie public interest reporting.

Approaches

AI strategies, laws, and policy proposals in the region generally address data protection and privacy in three ways. 

1. Establishing a data governance framework. Sri Lanka’s National Strategy on AI lays out how the country will develop and implement a new data strategy that supports Sri Lanka’s AI initiatives. The strategy will focus on three primary objectives: leveraging data as an asset, ensuring responsible data practices and promoting collaboration. To achieve this, Sri Lanka will establish a data governance framework to ensure compliance with relevant and upcoming laws and regulations, such as Sri Lanka’s Personal Data Protection Act (2022), the Right to Information Act (2016), the Electronic Transactions Act (2006), the forthcoming Cyber Security Bill and amendments to the Sri Lanka Telecommunications Act. The data governance framework will seek to achieve three key objectives: ensure relevant data is collected, standardized and accessible to appropriate stakeholders; guarantee the correctness and quality of the available data and developing processes for continuous assessment and improvement of data quality; and address ethical considerations related to data collection, storage and use. 
2. Strengthening data protection. Pakistan’s Regulation of Artificial Intelligence Act, for example, calls on the national AI commission to “strengthen telecommunication networks, digital infrastructure, data governance, data protection, and cybersecurity,” though it does not specify how it should do so. 
3. Incorporating privacy and data security into the AI lifecycle. Bangladesh’s National Artificial Intelligence Policy outlines how privacy should be incorporated into the development and use of AI technologies, stating that “models shall be trained with minimal use of potentially sensitive or personal data; personal data usage will require valid consent, notice, and the option to revoke; privacy measures like encryption, anonymization, and aggregation will be applied; [and] mechanisms shall allow users to flag privacy and data protection issues during data collection and processing.” Likewise, Bangladesh’s policy explains that the vulnerability of AI systems should be addressed through rigorous assessments and data validation, governance procedures and response protocols.

Impacts on journalism

While the journalism sector is not directly mentioned in any of the proposals, it will likely be impacted by them. As countries throughout South Asia further develop their data governance frameworks, newsrooms will be required to comply with these local regulations to ensure data privacy is built into any AI systems they use or deploy. Such data protection can help protect their reporters and sources. 

As in other regions, though, data protection laws can be misused to stifle public interest reporting. As such, policymakers should consider the impact their AI proposals can have on the news industry and consider exceptions within the laws to protect journalists’ ability to carry out investigations — as the Press Club of India and 21 media organizations have asked policymakers to do in response to India’s 2023 Digital Personal Data Protection Act. Though this law does not specifically address AI directly and, thus, was not part of our data set, it is worth highlighting since it is in the process of being operationalized and because of the significant press freedom concerns it has raised.

Public information and awareness

AI summary: Countries in South Asia are working with global groups to develop AI policies and are starting to train workers, establish school programs and launch public campaigns to raise awareness about artificial intelligence. These efforts could help journalists by giving them the skills to better understand and report on AI, even though the plans do not mention them directly.

Approaches

South Asia is drawing inspiration and financial support from global bodies such as UNESCO as countries seek to develop AI policies, though at this stage, AI literacy appears to be more of an update to existing curriculum rather than a standalone initiative. 

1. Training the workforce. Bangladesh’s National Artificial Intelligence Policy, for example, includes an objective to carry out “tailored training and skills programs [that] will address the AI skills gap in the workforce.”
2. Establishing AI programs with educational institutions. Bangladesh’s National Artificial Intelligence Policy further lays out a plan to carry out national campaigns targeted at educational institutions to encourage them “to establish and facilitate AI and data clubs.” Furthermore, Sri Lanka’s 2024 National Budget specifically allocates funding for new AI degree programs at state universities, in line with the country’s AI strategy. 
3. Raising public awareness. Sri Lanka’s National Strategy on AI lays out how the country will implement a public awareness campaign on AI: “We will develop engaging educational content on AI in Sinhala and Tamil for dissemination through mass media channels, such as television, radio, and social media. This content will explain key AI concepts, highlight potential benefits and risks, and emphasize the importance of AI skills in an increasingly digital future. By leveraging popular media and through collaborations with the private sector, we will reach a broad cross-section of society and foster a baseline understanding of AI.” Similarly, Pakistan’s proposed Regulation of Artificial Intelligence Act, which would establish a national AI commission, calls on the commission to “take proactive steps to create public awareness for positive and productive usage of Artificial Intelligence technologies for the benefits of the people.” 

Impacts on journalism

These initial developments, especially the support and best practices shared by international organizations such as UNESCO, are promising for the media sector. Even though they do not explicitly mention a role for journalists, they include provisions that could empower journalists to critically evaluate media content and could foster an informed society by advocating for greater information literacy among the public. Programs to train the workforce and students on AI, as well as public information campaigns, could benefit journalists — especially if some of these programs specifically target professional journalists and student journalists. This education would allow journalists to engage with AI for their own work, as well as to better understand and better report on AI. 

Conclusion

South Asia is one of the world’s most populous regions and home to rapidly advancing economies. Combine that with linguistic diversity and rapid adoption and proliferation of technology, and you have a group of countries whose AI policies will have far-reaching implications for journalism and beyond. As policymakers in South Asia think about how best to regulate AI, they should not only consider how these technologies can impact their economies, but also how these technologies can impact their information ecosystems, and the information ecosystems of their neighboring countries — both positively and negatively. Policies do not always need to mention journalism by name, but policymakers should still form working groups to assess how their proposals might affect journalism and determine how best to prevent negative consequences before these regulations are implemented.

The post South Asia appeared first on Center for News, Technology & Innovation.

The post Appendix appeared first on Center for News, Technology & Innovation.

On October 28 and 29, 2025, the Center for News, Technology and Innovation and Thomson Media brought together more than 35 journalists, product specialists and newsroom leaders from the Western Balkans and Central Europe in Sarajevo, Bosnia and Herzegovina, for “Journalism and AI: Building Resilient Newsrooms for the Future” — the second collaboration between the organizations. Over the course of two days, participants shared their struggles and successes integrating AI into their work, took master classes on AI prompt engineering and contemplated the question, “How do we harness AI’s power to build a strong future for our work?”

The answer that emerged was both multifaceted and promising. The journalists in this region are leaning in and figuring out what works best and what does not work for them and their audiences, and what they can learn from each other — and they often are managing with only small budgets and staff sizes.  

At one point during the workshop, CNTI Chair Craig Forman explained, “I wrote for the Wall Street Journal when I was a foreign correspondent, 30-ish years ago on the eve of the breakup of Yugoslavia.” He continued, “I bring that up, not to bring any ill will or bad memories, but to say that when I [was reporting in the region], it was unthinkable that we might be here today. Unthinkable. And as you all know, we have to think about the unthinkable, not only in the bad way, but in the good way.”

Forman’s words resonated deeply in a city once synonymous with war reporting but now hosting a discussion about how technology could safeguard truth and foster resilience in newsrooms.

AI Will Not Take Journalists’ Jobs, But Someone Who Uses AI Will

David Caswell, drawing on years of experience developing AI storytelling systems, kicked off the workshop with a powerful and straightforward insight: “We should take AI seriously because the trajectory of improvements has been radical.” He continued, “The effect of [AI investment] will be significant.”

This isn’t about jumping on every technological bandwagon or tacking the latest tool onto one’s current product. It’s about recognizing that structured storytelling models and AI-assisted workflows represent a practical evolution in how newsrooms operate. The question isn’t whether AI will change journalism, but whether journalism will adapt intelligently.

As Niamh Burns from Enders Analysis said, “AI will undoubtedly change how news is created, distributed, noticed and funded. This is really a moment where you [journalists] should all be thinking about what your value add is.”

These aren’t abstract problems for newsrooms in Sarajevo, Belgrade or Pristina. They’re immediate challenges that require practical solutions.

Sami Kçiku, a project manager at the independent news company Koha Group in Kosovo, shared that his newsroom was initially fearful of AI. Several other participants voiced similar stories of journalists worried they would be replaced by AI and, therefore, reluctant to try out new technology, with generational divides also at play. 

Kçiku offered the simple yet disarming advice that he gave to his team: “You won’t be replaced by AI, but you will be replaced by someone who uses AI.” Koha has since implemented a custom GPT to improve SEO and social media practices, and they use an external AI tool for transcription.

Journalists Must Begin to See Themselves as Innovators

As newsrooms attempt to address the tension between embracing AI and fearing it, Caswell recommended a calm but intentional approach of “innovation, adoption, diffusion.” 

It is important that innovation comes from collaboration between the editorial and product teams, and, as Burns pointed out, innovation should prioritize measurable success instead of bandwagon adoption. Nikola Bačić, editor-in-chief at Hercegovina Info, shared that his team has one AI meeting every week, where the editorial staff meets with the IT staff to discuss needs and potential solutions. 

Once a tool is created, newsrooms must take the time to properly train the entire staff on a technology to ensure they have the skills needed to adopt the AI tool, if they so choose. Tatjana Sekulic, an executive multimedia producer at N1 in Bosnia and Herzegovina, told the group that uptake at her outlet is mixed; some journalists are completely against AI, while others are overly reliant on it. As a result, the newsroom has implemented training for everyone to ensure they all have the skills to use AI responsibly. 

“It’s very important to teach them how to use AI in the proper way,” Sekulic said. “We’re investing in our knowledge and our people.”

Finally, following implementation, it is important to continue conversations to diffuse broader adoption of the technology. Veronika Munk, director of innovations at Denník N in Slovakia, explained that while full newsroom training sessions did not always achieve the desired impact for her team, they still provided useful insights. Over time, the organization found that complementing these larger trainings with a more targeted approach worked better. When Denník N introduces a new tool, they now focus on training smaller groups within the team or on one-on-one micro-trainings. These trained editors and reporters can then gradually share their knowledge with other colleagues. She also emphasized that the organization’s use of AI always involves human oversight.

While participants responded to this discussion with enthusiasm, some relayed concerns about the cost of implementing AI tools in newsrooms with already thin budgets. 

Damjan Dano, a tech entrepreneur from North Macedonia, explained that creating customized AI tools is not the only option; instead, there are a plethora of existing, inexpensive AI tools that newsrooms can use to support their work. Dano led participants through an exercise where they laid out their AI wish lists, and he shared a variety of existing tools that could be a solution to some of their needs. Frase, for example, can help with SEO optimization, and Asana AI can help manage newsroom workflows.  

“AI is a great tool, not your substitution,” he reminded participants. “Your job is safe, but you have to use the tools that exist today.”

These discussions came with a crucial caveat: Journalism’s future depends on “adaptability, but not at the expense of ethics,” Caro Kriel, chief executive of Thomson Foundation, said.

Ethics and Audience Relations Must Be a Core Part of Newsrooms’ AI Strategies

Another key theme over the two days of discussions was that AI should augment human judgment, not replace it. Newsrooms need to ensure their use of AI is driven by the values and ethics they espouse. It should enable and empower journalists to do their best work, to report relevant, important news and to deliver it to audiences effectively.

Marius Dragomir, director of the Media and Journalism Research Center, reminded participants that technology questions are never just technical. They’re about values, accountability and the social contract between journalists and their audiences. Every AI implementation carries an ethical weight that newsrooms must acknowledge and address.

“The trust of our audience is our highest value. To preserve this trust, we must be honest and transparent,” Vesna Ivanovska-Ilievska, co-founder and editor-in-chief of Umno.mk, said. 

In a session on AI, ethics and trust, Branislava Lovre, co-founder of AImpactful, noted that newsrooms can increase audience trust by communicating about their uses of AI clearly and effectively. 

As CNTI’s global AI Research Working Group has written, there is not one rule book or exact labeling technique that newsrooms should follow. Instead, audiences, like journalists, are still getting used to AI. What matters most is that journalists and their organizations are transparent about their use of AI and carry out a dialogue with audiences about what that means.

It is not enough to simply have an AI policy or guideline, Erjon Curraj, a digital transformation expert from Albania, cautioned. Newsrooms must implement these policies consistently and ensure their staff is aware of them, too.

“This isn’t just about disclosure,” Lovre concluded. “This is our unique chance to lead by example. Social media networks are overwhelmed by AI-generated content, and if we don’t try to explain to our audience what is happening, we won’t be in a good place in one or two years.” 

Journalism Producers Need to Pay Close Attention to Public Policy Related to AI

Participants also discussed how policies and government regulations can impact media freedom and ethics. 

Ana Toskić from Partners Serbia and Emily Wright from CNTI highlighted the potential impacts of EU digital laws, especially the AI Act and the Digital Services Act, on journalism in the Western Balkans. They discussed how in closed or partially closed media environments, where journalists rely on social media platforms to share stories, laws like the Digital Services Act will have a significant impact, especially when regulatory bodies do not operate separately from the government and can misuse the laws to stifle independent reporting. 

The pair further explained that media organizations located within the EU, and those in EU candidate states, will have to adhere to the General Data Protection Regulation (GDPR) and the AI Act when using AI in their operations. 

Gábor Kardos, CEO of the Hungarian publisher Magyar Jeti Zrt., emphasized the importance of staying on top of regulatory activity, especially in autocratic-leaning countries where independent media is the minority amongst state-controlled media. 

“Even if these regulations were created perfectly…. the state will always have the power to abuse them. And that’s the reason I’m advocating for better regulation,” Kardos said. “But we, as publishers, need to be aware that that’s not the thing that’s going to protect us. We as a community have to protect each other and ourselves, and be innovative and be faster than regulation can ever be.” 

Journalists Can Use AI to Help Address Challenges in the Information Environment 

In an era where press freedom is at risk, synthetic content floods our feeds and deepfakes grow more convincing by the day, journalists face an unprecedented challenge. Journalists are not just competing for attention anymore; they’re fighting for the very concept of verifiable truth.

The twist is that while AI can exacerbate some of the problems, it can also be part of the solution. Workshop participants recognized that AI tools could help newsrooms verify information faster and more thoroughly than ever before, allow them to reach new audiences and help them create new forms of content. The key lies in thinking ahead. 

“If we want to think strategically, we cannot focus only on the shortest term,” Kardos said. “It [the short term] does not matter if in five or seven years, it’s not journalism or AI, it’s humanity that will be in question. We need to focus on what happens in the midterm, within a few years.” 

Participants from several countries offered case studies of creative ways they have implemented AI into their workstreams, even on small teams with limited budgets that face pressure from their governments. 

The Center for Investigative Journalism of Serbia, which has a team of only 10 people, created a custom large language model (LLM) to analyze nearly 10 million pieces of data on wait times in Serbia’s healthcare system. The AI system allowed Ivana Milosavljević and her fellow reporters to analyze large amounts of data in record time. She noted, importantly, that a human reviewed all outputs to ensure accuracy, which was time-consuming, but she shared that AI enabled the team to reach and visualize conclusions in new and efficient ways. Initially tested on publicly available data, Milosavljević said the LLM system will be especially valuable for analyzing confidential data.

Ján Trangel, AI implementation lead at Ringier Slovakia, explained that the outlet has created an AI-driven hate speech moderation system after finding third-party tools ineffective for their language and regional needs. Ringier built custom hardware and an offline LLM — drawing on open-source options like Mistral and Google models — to evaluate messages in real time, classify their severity and support fully customizable moderation policies. The system processes more than 300,000 messages daily, reducing manual review while offering an admin panel for managing labels and decisions. Its offline architecture lets the team experiment freely, compare models and tailor features without depending on external cloud services. 

Vidi Vaka, a Skopje-based outlet with only three full-time reporters, created “KiberFlow,” an AI coworker that transforms the outlet’s reported stories into rap-style videos. KiberFlow uses character-animated performance and social satire to highlight everyday societal problems and has more than 1,000 followers on Instagram. The project is allowing Vidi Vaka to reach a younger audience that usually avoids traditional media in a new and engaging way. 

“For us, AI is not a shortcut; it’s a collaboration,” journalist Angela Petrovska said. “People follow KiberFlow not because it’s AI, but because it tells real stories made by good journalists.”

At Dennik N in Slovakia, Munk and her team explore AI uses on a smaller hiking website they manage. Every Thursday, the team publishes a set of recommended weekend hikes, using AI to assist with suggestions based on weather forecasts and difficulty levels. They have also developed tools for automated image cropping and social media posting. While not all types of tools are tested there, the hiking site provides an environment for experimenting, learning and identifying what might be useful before expanding these solutions to larger platforms at the different outlets in Denník N network.

Participants explained that AI tools are often easy to learn and improve creative flow, but they noted that human editing, cultural context and emotional nuance remain essential to keep the work authentic. Overall, they expressed optimism about the opportunities that AI tools can provide their newsrooms and left the workshop feeling motivated to see themselves as innovators in the news industry.

Conclusion

There’s something fitting about having this conversation in Sarajevo, a city that knows something about resilience in the face of existential challenges. The participants didn’t offer easy answers or technological determinism. Instead, they charted a middle path, one that takes AI seriously without surrendering the core values that make journalism essential.

The future won’t be built by those who reject AI wholesale or embrace it uncritically. It will be built by newsrooms that approach these tools with clear eyes, strong ethics and an unwavering commitment to serving their audiences with verified, trustworthy information.

The post AI for Sustainability: Building Journalism’s Future appeared first on Center for News, Technology & Innovation.

The post How Serbia could use EU Digital Services Act for state censorship appeared first on Center for News, Technology & Innovation.

— Amy Mitchell, Executive Director, CNTI

Read full story

The post Bridging Divides: Amy Mitchell on Shaping Journalism & Media Policy for the Digital Age appeared first on Center for News, Technology & Innovation.