The EBU has called for AI governance frameworks that recognize public service media as anchors of trust. It also demands concrete, and enforceable accountability from AI companies on the accuracy of AI-generated news, who sets the rules for trust online, and on the safety of children using AI systems, as the EBU has claimed in a statement.

The message was delivered alongside Fondation Abeona, the Global Trust Challenge, and 5Rights Foundation at a side event of the UN Global Dialogue on AI Governance, in Geneva.

“AI companies are inserting themselves between newsrooms and audiences, making journalism accountability methods such as bylines and editorial principles invisible to audiences, preventing trust from being maintained and increased”, explained Luis Jiménez, Head of Business Development and Partnerships for Eurovision News (EBU).

The warning comes as public trust in news and reliance on unverified AI tools move in opposite directions. According to the Reuters Institute’s Digital News Report 2026, overall news trust has fallen to 37%, its lowest level since 2015, while AI chatbot use for news has reached 10% of respondents and continues to grow.

A joint EBU/BBC study, News Integrity in AI Assistants (October 2025), found that 45% of AI assistant responses tested had significant issues overall, with 31% showing sourcing problems and 20% showing accuracy issues.

Trust, panellists argued, can’t be built by any single actor. Tanya Perelmuter, of Fondation Abeona and Ambassador for the Global Trust Challenge, remarked: “Trust isn’t just a principle, it’s an engineering challenge, a policy challenge, and a global collaboration challenge. It means knowing the information we rely on is authentic, traceable, and aligned with shared standards of integrity. Trust architecture is a public good that cannot be outsourced to private, non-transparent players who were not democratically elected. No single organization can solve this alone. Grassroots, bottom-up collaboration is necessary”.

Speakers also raised the disproportionate risks posed to children by generative AI. Marie-Ève Nadeau of 5Rights Foundation said children are adopting generative AI three times faster than adults yet are an afterthought in its design. According to the OECD, AI-related incident reports concerning child safety doubled between 2022 and 2025.

“Time and again we have seen the consequences when growth and profit are put ahead of children’s safety, and when tech firms operate without accountability”, Nadeau added. “It doesn’t have to be this way. If we ensure AI only reaches children when it is demonstrated to be safe, accurate, and supportive of their healthy development, children can also benefit from the possibilities offered by AI”.

Keita Azuma, Policy Researcher at UNU/UNITAR, placed these concerns in a wider context, noting that AI governance is entering a new phase, from regulating future risks to building the capacity of societies to manage risks already reshaping the information environment. Multilateral cooperation, he said, should strengthen AI literacy, societal resilience, and interoperable governance, because trustworthy AI ultimately depends on trustworthy societies.

Building on this discussion, the organizations put forward three asks to the UN Global Dialogue on AI Governance:

1. That the Dialogue recognize public service media’s role as a trust anchor within national AI governance frameworks (EBU).
2. That AI systems affecting children be demonstrably safe, accurate, and effective before reaching market, with the burden of proof placed on developers and deployers (5Rights Foundation, backed by 100+ organisations and experts).
3. That open, interoperable standards and sandboxing environments be developed collaboratively by innovators and public officials so that information infrastructure can’t be unilaterally dictated by technology companies (Fondation Abeona / Global Trust Challenge).

As they stated, AI reshapes how audiences find and evaluate information, public service media’s role as an independent, accountable source of verified news becomes more important, not less. Embedding that principle into AI governance frameworks is essential to keeping trust achievable in the AI era.

The Global Trust Challenge

The Global Trust Challenge is a global coalition, including IEEE SA, OECD-GPAI, AI Commons, UNESCO, the World Bank, and IDB, bringing together expertise from various sectors and regions.

It is a worldwide initiative that seeks interdisciplinary solutions combining technology and policy to address challenges posed by generative AI, aiming to develop innovative approaches for ensuring the verification and trustworthiness of AI-generated content.

Fondation Abeona

Fondation Abeona guides citizens, companies, and public decision-makers in understanding and using AI, and addresses the societal impacts associated with its deployment, partnering with key actors worldwide.

5Rights Foundation

5Rights Foundation is an international NGO working with and for children to build a digital world that honours their rights as set out in the UN Convention on the Rights of the Child.

UN Initiatives for education and training on AI capacity building

Through research, training, and knowledge-sharing initiatives, UNU and UNITAR are building AI governance capacity by equipping policymakers, public officials, and other stakeholders with practical skills for the safe, secure, and trustworthy use of AI.