EU Statement – UN General Assembly: Informal Consultations on Artificial Intelligence

14 March 2025, New York – Statement on behalf of the European Union and its Member States delivered by the European Union Delegation to the United Nations at the United Nations Informal Consultations on Artificial Intelligence

Check against delivery


 

 

Excellencies, Co-facilitators, dear colleagues,

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

The Candidate Countries Türkiye, North Macedonia*, Montenegro*, Serbia*, Albania*, Ukraine, the Republic of Moldova, Bosnia and Herzegovina* and Georgia, and the EFTA country Norway, member of the European Economic Area, align themselves with this statement.

Let me first thank the co-facilitators for the elements paper which provides a solid basis for our discussion today and helps frame an operational discussion. It is promising to see a number of elements where the views of the membership are converging. The EU remains committed to shaping an effective, agile and lean multi-stakeholder process for AI governance at the UN level in line with the GDC.

Today we would like to highlight the aspects we believe we can already build on further, as well as those that require more in-depth discussions.

First, on the panel. 

We welcome the mandate of the panel as presented in the elements paper, as a scientific multidisciplinary body enjoying full independence in terms of work plan and outputs. We nonetheless reiterate that the panel’s mandate should be focused on providing the state of science on artificial intelligence, by credibly synthesizing and consolidating the available research and evidence, rather than initiating new research. It should also build on existing national, regional and international initiatives and research networks, without duplicating them.

On the panel’s structure, we support in general the elements put forward in the paper. With regards to the selection of panel members specifically, we call for a fully transparent selection process conducted by an independent committee on the basis of scientific merit and expertise as well as criteria defined by Member States to ensure inclusivity and representativeness of scientific experts from a geographical and gender balance perspective. Specifically, adequate representation of experts from the Global South to address the specific challenges and needs of these regions should be ensured. The structure of the panel itself should be light and agile, in order to be set up quickly and to respond to the fast-paced development of the technology.While fully supportive of the IPCC as a credible scientific body supporting policy discussions on climate change, we do not think it is replicable to our current discussions on AI. The imperatives of independence and agility call for a smaller yet representative body, entrusted with the issuance of scientific assessments at a fast pace. 

Finally, the panel outputs as presented in the paper seem overall appropriate. The panel should aim to produce annual reports which should feed into and be discussed in existing processes, including the Global Dialogue, as well as concise briefs on relevant developments.

Second, on the dialogue.

We welcome the co-facs’ proposal to have a multistakeholder platform to facilitate open, transparent and inclusive discussions on AI governance. We suggest reinforcing the mandate by framing it around how AI can serve the public interest by contributing to the achievement of the SDGs, reducing the global digital divides and strengthening the respect for, promotion and protection of human rights, as well as concrete key factors of success to achieve this goal. Besides, it could identify meaningful approaches to progress on questions of international cooperation, safety, security and trustworthiness. During the GDC negotiations we also agreed that creating an additional mechanism should not come at the cost of efficiency and coherence. To mandate the Dialogue to work in synergy with existing initiatives is therefore crucial. We are however against creating a hierarchy of meetings which could be detrimental to pre-existing initiatives, and leave it to the dialogue’s secretariat to best ensure those synergies. While the dialogue should create synergies within the UN, we would benefit from clarification on what is meant by combining existing meetings and fora. 

On the structure, the Dialogue being conceived as a multistakeholder platform, it would not make sense to have two separate tracks for governments and other stakeholders. The added-value of the Dialogue will lie in its ability to create a shared space for all stakeholders to interact, therefore it should only have one track. At this stage, an annual meeting seems like a maximum reasonable commitment, as a higher frequency will undoubtedly pose logistical and financial challenges, and create a burden on smaller delegations. To mitigate such logistical and financial challenges, the Global Dialogue should take place in the margins of existing United Nations multi-stakeholder conferences and meetings, and allow for hybrid participation. It is also our preference to link the annual dialogue to the publication of the panel’s annual report. We note that New York is mentioned as a location of relevant UN conferences and meetings, and reiterate that there are three conferences the EU sees as preferential because of their lower barriers for participation: the AI4Good summit, the IGF, and UNESCO’s Global Forum on the Ethics of AI. None of these currently take place in New York.

We agree that the dialogue’s outputs should be non-binding. We also reiterate they should be non-negotiated outcomes, in order to avoid lengthy high-level negotiations taking over the more detailed substantive discussions that should be at the core of the Dialogue. The dialogue’s outputs should therefore consist of summaries of discussions, including the contributions of civil society and other non-state stakeholders, prepared by the secretariat. The outputs of the dialogue should in no way dictate the work of the panel as the EU sees the dialogue and the panel as institutionally and functionally separate. Should aspects that have not been covered in the panel’s work emerge during the dialogue, the independent panel may choose to consider them if deemed relevant and important. The panel would not be obliged to take into consideration the inputs from the discussions taking place during the Dialogue or any other existing process.

Once we have worked out the details of the two mechanisms, we should then dedicate some time to discussing what the secretariats that will support them should look like in light of the budgetary implications.

Our initial thinking is that there should be two separate lean secretariats for the panel and the dialogue. The panel secretariat should be free from political influence, and provide substantive support to the panel members in preparing the outputs throughout the year; it might therefore be necessary to have full-time dedicated secretariat members to that end. Such secretariat could be coordinated and guided by the Inter-Agency Working Group on AI. For the dialogue, a leaner secretariat might be sufficient to organise the annual gathering and prepare a summary of discussions.

Colleagues, 

We need to keep in mind the dire budgetary situation in which the UN finds itself, avoiding new mandates in the current budget cycle when possible, and respecting the UN’s rules and procedures, especially rule 153. If we want the panel and dialogue to deliver, we need to realistically explore all financing options, including voluntary contributions.

We look forward to hearing other delegations’ views on those elements. The EU remains committed to engaging constructively with all stakeholders throughout this process. 

Thank you.


 

* North Macedonia, Montenegro, Serbia, Albania and Bosnia and Herzegovina continue to be part of the Stabilisation and Association Process.