EU Statement Global Dialogue on AI Governance: Informal thematic deep-dive session

PRINT

EU Statement -- Global Dialogue on AI Governance: Informal thematic deep-dive session

14.05.2026 Press and information team of the Delegation to the UN in New York

14 May 2026, New York -- Statement on behalf of the European Union and its Member States delivered by Ms. Daphné Barbotte, EU Delegation to the UN, at the Global Dialogue on AI Governance: Informal thematic deep-dive session

Excellencies, dear co-chairs,

Dear colleagues,

I have the honour to deliver this statement on behalf of the European Union and its Member States.

We thank once again the co-Chairs for the leadership and the extensive consultations. Building on last week's meeting, we are pleased to share our views today on the content of the discussions we hope for at the Dialogue.

1. On the different clusters:

Cluster 1: AI opportunities and implications: social, economic, ethical, cultural, linguistic and technical dimensions

The Global Dialogue is the appropriate forum to gather all stakeholders and exchange on the broader implications of AI on jobs, culture and creation, language diversity, economic fairness as well as environment and climate protection. The Dialogue should discuss how AI ecosystems can be inclusive and sustainable from the onset.

At the same time, the framing of the Dialogue's priorities should not overlook the positive transformative potential of AI as a force for societal good. AI is already driving innovation, efficiency, and public benefit across key sectors and holds significant promise for education, through personalised learning and improved access in underserved areas, as well as for the public sector more broadly. The Dialogue should ensure that responsible AI innovation is explicitly framed as an opportunity, not only as a risk to be managed.

In this field, the EU is already setting up an AI Observatory to provide robust indicators to assess the impact of AI in priority sectors, monitor developments and trends, and the changes it may bring to the labour market - as this is of particular interest to us.

The EU itself is made up of a vibrant linguistic and cultural diversity. AI systems trained predominantly on data from a limited number of languages produce less accurate and more costly outputs for speakers of underrepresented languages. This raises important questions about equitable access to and the fair distribution of value derived from linguistic and cultural resources, which the Dialogue should address as a cross-cutting governance priority.

We want to ensure AI-driven growth benefits all. This requires discussion of who controls the data, the compute, and the value chains that underpin AI systems today.
The Dialogue should also surface the question of equitable value distribution from AI systems built on public data, linguistic and cultural resources, and collectively generated knowledge.

Moreover, we strongly believe that addressing the environmental and climate impacts and opportunities of AI is a key part of the Dialogue's general SDG-oriented mandate and should be addressed accordingly.

Finally, data governance as a strategic public good deserves explicit recognition under this cluster. Global asymmetries in access to and control over data are foundational to AI inequalities. Frameworks promoting data-sharing, data sovereignty, and public-interest data commons - particularly for developing countries - are essential to ensure that the value derived from data is distributed equitably and that all societies can meaningfully participate in AI development.

Cluster 2: Bridging AI divides: capacity-building, access and digital foundations

To ensure that individuals, organisations, and institutions can engage with AI confidently, critically, and safely, AI capacity-building, in the broadest sense, must be recognised as a cornerstone of effective AI governance - underpinning trustworthy AI, the protection of fundamental rights, and the interoperability and compatibility of governance frameworks. Without robust human and institutional capacity, even the most well-designed policies risk remaining declarative rather than operational.

A comprehensive approach is required to strengthen capacity at all levels, as well as globally, especially in developing countries, moving beyond basic AI literacy to foster informed agency - empowering all stakeholders to evaluate, question, and deploy AI systems responsibly.

As AI becomes increasingly embedded in society, capacity-building is not merely a standalone priority but a prerequisite for achieving meaningful and sustainable outcomes in all thematic areas of AI governance. By investing in skills, systems, and mindsets, we can ensure that AI is developed and deployed in ways that are responsible, inclusive, and aligned with societal values.

International cooperation is essential, and the existing ecosystem of UN initiatives alongside bodies such as the OECD, GPAI, and the Council of Europe, offers a foundation to build on. The Dialogue should map these resources, identify genuine gaps, and ask which mechanisms have delivered results and which have not.

In this context, strengthening linkages with existing multistakeholder processes is essential. These platforms provide structured and inclusive spaces for continuous dialogue, and can function as operational bridges between global governance discussions and national and local implementation experiences.

In particular, the EU has a lot to share with regards to open-source approaches in AI. They can enable more equitable participation in the development of meaningful AI applications as digital public goods, where stakeholders can actively engage in AI innovation and oversight.

A further dimension that must be addressed under this cluster is the emergence of AI as critical digital infrastructure. The Dialogue should treat AI infrastructure as a matter of public interest - not only of market access or proprietary control - and explore mechanisms to support developing countries in building sovereign capacity across the full AI stack.

Finally, the Dialogue should give explicit attention to agentic AI as an emerging governance frontier. Agentic systems raise new and distinct challenges around accountability, oversight, and control that existing frameworks do not yet adequately address. A forward-looking capacity-building agenda must prepare institutions to govern not only current AI applications but these rapidly evolving paradigms.

Cluster 3: Safe, secure and trustworthy AI: interoperability and compatibility of approaches

Interoperability and compatibility of governance approaches should be one of the main objectives for the Global Dialogue. Ensuring a certain level of compatibility of approaches building on existing binding frameworks would help prevent regulatory arbitrage between jurisdictions, especially with regards to systemic and universal risks, while fostering the diffusion of good practices and facilitating responsible cross-border deployment of AI.

Promoting and implementing our collective engagement towards international law and human rights online is the common minimum base in this regard. Discussions on this topic could ideally pave the way for a stronger articulation of the multitude of international initiatives, with a view to better leveraging their respective strengths and complementarities in a coordinated manner, based on shared principles and approaches.

While we recognise that different legal traditions, development contexts, and societal values will produce different governance choices, we would advocate for a baseline of shared binding principles and guardrails, notably international human rights law, that can function as a common reference point across diverse frameworks.

We see cluster 3 and 4 as very interlinked. Safe, secure and trustworthy AI is essential to prevent harm from unreliable, malicious systems or misuse. For this we need transparency, accuracy, information integrity, cybersecurity, and accountability so citizens and businesses can trust and confidently adopt AI technologies.

Without transparency, accountability, and human oversight, safe, secure and trustworthy AI will remain an aspiration rather than a realistic AI governance goal. Equally these criteria are a precondition for testing AI systems for human rights compliance and assessing their implications for societies.

Beyond formal compliance, the Dialogue should promote a shift towards verifiable trust in AI systems. Trust cannot rest on self-assessment or voluntary disclosure alone. It requires the ability to independently assess systems through auditability, traceability, and validation mechanisms across the AI lifecycle - drawing on established practices from cybersecurity, safety engineering, and risk assurance. Convergence on practical assurance mechanisms, including shared audit methodologies, certification approaches, and evaluation frameworks for high-risk AI systems, would be a concrete and high-value output of the Dialogue.

Cluster 4: Respecting, protecting and promoting human rights: transparency, accountability and human oversight

AI systems must be designed, developed and deployed in ways that respect and promote dignity, equality, and human rights, as well as democratic principles and processes so that users can continue to benefit from AI tools and are protected from abuses and violations. This includes information integrity, and mitigating risks related to AI-enabled manipulation of public discourse. The EU therefore advocates for a human-centric and human rights-based approach to AI, premised on international law, in particular international humanitarian and human rights law.

Online protection of women and girls, who are disproportionately targeted by AI-facilitated gender-based violence and other harmful practices, should be an absolute priority in this regard, with the active involvement of UN Women and OHCHR. Likewise, the protection of children and older persons against harmful AI-generated content, manipulative algorithmic systems, and data exploitation, requires a dedicated response and should be addressed as a separate, frontline issue, including through the implication of relevant UN agencies (UNESCO, UN Women, OHCHR etc.).

In addition, AI-enabled surveillance technologies pose acute risks to freedom of opinion and expression, freedom of assembly, and safety of journalists, human rights defenders, and marginalised communities. The findings of the UN Special Rapporteur on freedom of opinion and expression, and OHCHR's work on surveillance technologies, provide directly relevant frameworks that this Dialogue should engage with explicitly.

This being said, we have reservations about consolidating human rights with transparency, accountability and human oversight on a same footing. These themes raise questions of a different nature. Human rights in the context of AI engage binding obligations under international human rights law and concern the full spectrum of rights affected throughout the AI lifecycle, while transparency, accountability and human oversight refer primarily to procedural and technical safeguards which should primarily help enforce human rights commitments. This could be better reflected in the proposed clustering.

Now allow me to switch to Spanish for this next point.

1. On existing policies, approaches and best practices:

La UE está dispuesta a compartir su experiencia en la promoción de un enfoque de innovación responsable en materia de IA.

La IA es una poderosa fuerza para el cambio positivo en toda Europa, impulsando la innovación, la eficiencia y el beneficio social. Al fomentar la innovación responsable en materia de IA y una política que antepone la IA, la UE está liberando el potencial transformador en sectores clave. Más allá de la industria, la UE está democratizando sus beneficios mediante la integración de la IA generativa en el sector público, garantizando que estas tecnologías sean coherentes con los valores europeos.

La UE también tiene una experiencia sustancial en la armonización de los enfoques políticos en todos los Estados miembros de la UE en diversos ámbitos digitales y en la superación de los retos de gobernanza en el ámbito de la IA. La UE está dispuesta a compartir su experiencia con otros países a través del Diálogo global.

Acogeríamos con especial satisfacción las contribuciones de los organismos regionales y los países que han desarrollado enfoques de gobernanza fuera del marco de la UE y la OCDE. La estrategia continental de IA de la Unión Africana, los marcos de gobernanza de la IA de la ASEAN y las experiencias nacionales de los países en desarrollo representan el conocimiento que este Diálogo necesita.

La actual arquitectura de gobernanza de la IA de las Naciones Unidas, que incluye la UNESCO, la UIT, el PNUD, el ACNUDH y otros, junto con la GPAI, el Consejo de Europa, el Proceso de Hiroshima y la serie de Cumbres sobre la IA, proporcionan un nutrido elenco de referencias. El Diálogo debe evaluar este panorama con honestidad: qué mecanismos han funcionado, cuáles han duplicado el esfuerzo y dónde siguen existiendo lagunas reales.

El Diálogo podría examinar las sinergias con las organizaciones e iniciativas internacionales, con el objetivo de conectar los esfuerzos, tender puentes en interoperabilidad y enfoques compartidos. El Diálogo puede servir de puente y mecanismo para traducir los importantes principios y marcos elaborados por los diversos órganos descritos anteriormente en compromisos viables para los miembros.

Permítanme cambiar ahora al francés.

1. Concernant la continuité entre les Dialogues :

Le succès du dialogue dépendra de sa capacité à aller au-delà des sessions plénières traditionnelles des Nations unies et à agir en tant que véritable plateforme multipartite. Nous nous attendons donc à ce que les sessions thématiques soient participatives et interactives, et qu'elles offrent un espace d'échanges ouverts entre tous les participants, en facilitant une rencontre entre l'« offre » et la « demande ». Les intervenants devraient être encouragés à demander l'avis du public participant au Dialogue sur des situations et des questions spécifiques, afin que les sessions thématiques contribuent à formuler des propositions concrètes et aboutissent à de nouveaux partenariats.

Afin d'assurer la continuité entre la première et la deuxième édition du Dialogue, en s'appuyant sur les mécanismes de coordination existants et sur l'initiative des coprésidents du Dialogue, des groupes informels auto-animés composés de tous les participants à une session thématique spécifique pourraient être mis en place, afin de permettre des discussions de suivi. Ces groupes pourraient également être ouverts aux parties prenantes intéressées qui n'ont pas été en mesure de participer à la session thématique initiale, afin qu'elles puissent enrichir les débats et en tirer parti.

Parallèlement aux prochains dialogues en présentiel, le dialogue pourrait également envisager d'intégrer des consultations thématiques régulières organisées autour des thèmes principaux du dialogue, qui alimenteraient les engagements en présentiel. Des consultations ouvertes à tous les États membres et à toutes les parties prenantes contribueraient à garantir une participation continue et significative, en s'appuyant sur la dynamique de juillet 2026 à mai 2027.

En outre, le dialogue pourrait envisager une feuille de route pour mettre ce forum au cœur d'une approche cohérente et participative de la gouvernance mondiale, et préciser la manière dont le dialogue facilite la complémentarité entre les forums et mécanismes existants.

1. Concernant le résumé des co-presidents :

Le résumé des coprésidents devrait fournir un aperçu précis de l'état d'avancement des discussions sur l'IA à l'ONU. Il devrait refléter à la fois les points de vue convergents et divergents des participants et servir de référentiel des meilleures pratiques et des solutions potentielles proposées au cours des deux jours.

Nous serions également favorables à une analyse structurée des domaines de gouvernance dans lesquels les cadres internationaux restent lacunaires ou inadéquats, et à une recommandation d'ordre du jour pour les travaux intersessionnels, afin que les discussions ultérieures des Nations unies sur la gouvernance de l'IA, y compris le deuxième dialogue mondial en 2027, s'appuient sur un précédent et ne repartent pas de zéro à chaque fois.

Je vous remercie.