As Artificial Intelligence (AI) is fundamentally reshaping our information ecosystems, public interest media faces historical challenges regarding its sustainability. Facilitated by the Forum on Information and Democracy and led by Luxembourg, Spain and Ukraine, the Workstream on Safeguarding Access to Reliable Information in the Age of AI held its second meeting highlighting this specific challenge and possible policy responses. 

On 18 June 2026, the Forum on Information and Democracy held the second meeting of its workstream on Safeguarding access to reliable information in the age of AI. On this occasion, María González Veracruz, Secretary of State for Digitalisation and Artificial Intelligence of Spain, announced that the country was joining Luxembourg and Ukraine in the co-chairship of the workstream. As she noted,

“False content is easy to spread and difficult to disprove…international cooperation is essential for the overarching goal of protecting information integrity, trust, and democracy.”

Gathering representatives of signatory States of the Partnership as well as civil society experts and researchers, the first part of the meeting highlighted the detrimental impact of AI on the economic model of media.

The second half of the meeting focused on policy responses to establish fair remuneration and licensing schemes, building on some approaches in the UK and in Italy. 

Camille Grenier, Executive Director of the Forum on Information and Democracy, highlighted the importance for the media community to be involved in AI Governance spaces.

“As discussions on AI Governance move forward, we believe that our workstream can play a meaningful role in elevating the specific impact of this technology on public interest media. This is the reason we are engaging with the UN AI Global Dialogue to be held in Geneva on 6-7 July.”

The next meeting of the Workstream, to be held in September 2026, will more specifically focus on epistemic risks due to AI and will base its discussion on the results of an international survey prepared by the Ministry of Culture of Ukraine and the FID. All stakeholders are invited to contribute to this work. 

Key Insights

How does AI affect fair remuneration to news media? 

• Evidence shows that AI-based summaries of the news reduce traffic, and hence revenue, to online news websites. 
• Big tech companies use exceptions in IP law, such as fair use and text-and-data mining exceptions, to ingest training data for their proprietary models. They then use the same copyright protections to secure the value they generate using public information. 
• Ad spending and algorithms decide what information is amplified, and hence monetized and funded online. This gradually disincentivises and marginalizes high-value, trustworthy news. 

Concrete cases to rebalance the power asymmetry

• Litigation on copyright: in Meta v. AGCOM, the Court of Justice of the European Union emphasised that publishers’ rights under the Copyright in the Digital Single Market Directive (CDSMD) are exclusive. 
• Competition law: In the UK, the Competition and Markets Authority imposed conduct requirements on Google, including to provide information on how publishers’ content is used to fuel generative AI features and ensure that content is attributed and links back to the publisher’s original sites. 

Both these cases point to an emerging trend of regulated contractual freedoms, where publisher rights are placed front and centre even as tech companies are free to use their content within their products. However, how remuneration must be calculated, and how bargaining must take place are still open questions.

The case of licensing regimes

Licensing is a much discussed solution and can be helpful in several contexts.  

However, specific attention should be paid to the following:  

• Larger outlets are in a unique bargaining position to gain licensing deals and safeguard their revenues, which puts smaller, more local news media outlets frequently serving underserved areas or ‘news deserts’ at a disadvantage. These media deals are overwhelmingly concentrated in the US and the EU, leaving the rest of the world far behind. 
• Those remedial policies focused exclusively on compensation, rights-holding and licensing to news media organizations will affect populations, overwhelmingly in the Global South, who cannot afford subscription fees and may be priced out of a reliable information ecosystem. 

What other solutions remain available? 

• Publishers can include ‘do not crawl’ signals in their work. 
• They can engage in individual licensing deals and collective bargaining in tandem – these are not mutually exclusive. 
• Working with national organizations pushing for policies and safeguards is also helpful. 
• Information integrity and transparency principles must be embedded in national law, regulation and multilateral agreements, and states must engage in international declarations and initiatives. 
• Policymakers must encourage building lightweight, local AI tools for underserved populations and languages, as opposed to viewing large-scale indigenous frontier models as the only solution. 
• Because AI’s future impact is uncertain, policy should not assume a single predetermined model.
• Press aid initiatives can be helpful to create conditions for independent journalists to produce reliable information, thus treating journalism as a public good. 
• Reliable journalistic content must be identifiable and trustworthy even when mediated through AI systems. 
• According to some experts, in certain contexts, the conversation could move beyond copyright and compensation towards the much larger goal of protecting knowledge infrastructures as part of cultural commons.
   

The first meeting of this workstream was held on 29 April, 2026. It gathered representatives of signatory States of the Partnership as well as civil society experts and researchers. The meeting involved expounding the current economic challenges to the information ecosystem in the face of AI. It also considered the existential threats to media sustainability, and what a weakened media ecosystem does to public trust and knowledge. Finally, the meeting briefly touched upon global regulatory approaches, and started discussions on possible policy interventions.

 

Key Insights

This first meeting revealed the multitude of challenges faced in the realm of AI and the information space globally.

Power Dynamics, Copyright and Remuneration

The prevalent use by AI chatbots of journalistic content reveals a shift in power dynamics. This affects the remuneration of media houses for their work – either in the form of training data, in the form of data that is extracted and summarised, or in terms of their revenue from ad traffic reduced by AI summary tools. There is also a severe lack of transparency in terms of what data is used, and how credit for this content is attributed to the press. This is also related to the problem of fair compensation to all parties in the AI value chain, from frontline journalists, to editors and even to data workers.

The first question, then, is what kind of copyright or licensing regime is best placed to put power back in the hands of information gatekeepers, viz. Media houses, for all these use cases. What kind of bargaining regime is best placed to do so? What would fostering transparency and provenance data look like, how do we ensure fair remuneration, and in what form must we consider digital levies and taxes?

Deepfakes and Epistemic Challenges

A second problem is the prevalence of hyperrealistic content, i.e. deepfakes and disinformation. Deepfakes are tied to large scale financial fraud, gendered violence, attacks on business ecosystems and public figures. Disinformation in sensitive contexts – such as war, elections and natural disasters – is especially worrying. These are further complicated by the use of agentic AI to scale such information attacks, including, but not limited to, Foreign Information Manipulation and Interference (FIMI).

The detection of deepfakes is increasingly difficult, and costs more than its generation. This is compounded by linguistic divides, regional variations, and a necessity to understand specific, local, contexts. This hampers detection, moderation and response – especially in the Global South – when AI tools are overwhelmingly developed and trained in the West.

The prevalence of deepfakes, particularly in the form of ‘replica websites’ of established public service media, affects the latter’s visibility, trustworthiness and reputation. This consequently affects search engine traffic and remuneration too. Sourcing errors and wrong answers from prevalent AI chatbots in summarizing the news – whether via a lack of context, mistaken attribution, lack of sourcing, editorialization or inaccuracy – also have these effects. This affects the perception of the media as a bulwark of truth and a reliable fourth pillar of democracies worldwide.

Where do we go from here?

Solving these issues would require fundamentally strengthening societal resilience via publication policies, but also risk-based governance, heightened safeguards, deployment of privacy law, data protection and targeted criminal law.

Taken together, these problems affect media pluralism; in the form of the media’s ability to report, especially from underserved and underreported regions, and in terms of visibility for the same. This affects access to reliable information for the general public. 

 

What do we want to achieve?

 

At our next meeting, we hope to present policy interventions using a comparative, global and analytical research approach. The next meeting is scheduled for the 18th of June, 2026. Civil society organizations are invited to share their work and policy interventions, while we also hope to learn from individual member states’ experiences. 

As generative AI poses new threats to the right to information and election integrity, the governments of Ukraine and Luxembourg announced the creation of a new workstream in the framework of the Partnership for Information and Democracy. This initiative was announced at the launch of the 2026 annual plan of action of the Partnership. 

On Thursday 19th, the Forum on Information and Democracy announced its annual plan of action during a meeting gathering representatives from governments, civil society and the research community. The plan comprises a series of international processes and meetings as well as ad hoc initiatives and projects.

“In a deteriorating global landscape for the regulation of our information ecosystems, we are proud to count on the support of our signatory States and community,” explains Camille Grenier, Executive Director. “The 2026 plan of action will lead to direct impact, ensuring best practices are shared amongst different stakeholders to harness the positive potential of technology while respecting democratic principles and human rights.”

Tackling the challenges of AI

As AI keeps posing new threats to information and election integrity, the plan notably includes the launch of a new Workstream on Safeguarding Reliable Information in the Age of AI led by the governments of Ukraine and Luxembourg.

Based on a thorough consultation of signatory States and partners from civil society and research, the workstream will focus on two pillars. First, it will address the disruptive role of AI in the news market and the sustainability of journalism. As AI exacerbates existing challenges in the media’s business model, the workstream will notably seek to counterbalance the power asymmetry between Big Tech and Public Interest Media. 

Second, the group will address the misuse of generative AI and its impact on information and election integrity. It will notably seek to create governance and accountability structures to counter the misuse of AI for information manipulation. 

From Belém to implementation 

The governments of Brazil and Armenia, on their part, will remain committed to their workstream on information integrity on climate change and the environment, launched in February 2025. 

The second phase of this workstream will notably include the mapping of policy and regulatory responses in line with the Bélem Declaration on Information Integrity on Climate Change, endorsed by more than 20 Signatory States of the PID, and the recently published 10 priorities to end climate and environmental disinformation. 

Through the workstream, FID will work with countries to develop concrete commitments ahead of COP17 on Biodiversity to be held in Armenia and COP31 to be held in Turkey.