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.

First of all, let me congratulate the Permanent Representatives of Costa Rica and Spain on their appointment as co-facilitators for this important process. Excellencies, thank you for convening this first consultation. The European Union is glad to see GDC implementation moving forward swiftly.

Artificial intelligence is a technology that inherently carries tremendous opportunities, but also risks. The discussions we are starting today, building on the adoption of the Global Digital Compact, the UNESCO Recommendation on the Ethics of AI, and other relevant texts adopted in the UN General Assembly, will contribute to defining the best ways in which AI can be put to the service of humanity, balancing regulation and innovation, while avoiding adverse effects on human rights, development and the environment.

The EU stands for a human-rights based, human-centric and multistakeholder approach to governing emerging technologies. We are ready to engage, collaborate, and share our experience stemming from the development of the EU AI Act and other instruments and programmes. We are also eager to learn from major upcoming events such as the Paris AI Action Summit and Kigali’s Global AI Summit in Africa that can feed our reflexion.

As a starting point, we are pleased to share today the EU’s initial views on the Independent International Scientific Panel on AI and the Global Dialogue on AI Governance.

Before outlining the specific points related to the panel and the dialogue, we would like to highlight several horizontal principles. It is important to i) maintain the panel and the dialogue as separate discussions, albeit linked, and to ensure a balance between New York, Geneva, and Paris; ii) ensure adequate focus on development objectives; iii) and strengthen the multi-stakeholder approach to international AI governance.

Both the Panel and Global Dialogue should, at their core, give due consideration to the promotion and use of local languages and content across AI, as a way for AI to be truly and meaningfully inclusive.

More specifically on the panel:

• It is of utmost importance that the new Panel integrates itself into the existing structures of the UN system and creates mutual synergies by benefitting from existing expertise in the UN system, from UNESCO, ITU, OHCHR and others.
• The panel should be fully independent and free from political influence. It should be composed of scientists, reflecting a balanced geographical and gender representation from a broad range of disciplines. The panel should be required to provide technical expertise to cover different aspects of AI opportunities and risks, including its political, social, economic, environmental and cultural impacts.
• Its objective should be to provide the state of science on artificial intelligence, by credibly synthesizing the available research and evidence on the state of technology development, as well as risks and opportunities emerging from the application of AI, drawing on existing national, regional and international initiatives and research networks.
• The panel should avoid prescribing policy, but rather focus on synthesising and consolidating existing AI research through evidence-based impact, risk and opportunity assessments, to provide policymakers the necessary elements for knowledge-based decisions and the promotion of scientific understanding among stakeholders. It can however formulate conclusions as assessed and considered necessary.
• The panel should aim to produce at minimum annual reports which could then be discussed in existing processes.

Second, on the dialogue:

• AI governance requires a holistic and global approach, where we should aim to turn a patchwork of initiatives into a coherent approach in compliance with international law, human rights and the SDGs.
• The Global Dialogue should be a structured and inclusive platform within the United Nations system, building upon existing initiatives and facilitating multi-stakeholder cooperation to share policy approaches to AI governance and best practices, while avoiding risks and negative impacts, and contribute to an understanding of capabilities, opportunities and risks of AI.
• It should also be a platform for discussion of practical strategies for harnessing global AI opportunities and addressing global AI challenges, governance and regulation, including the impacts of AI on human rights and fundamental freedoms, on democratic access, as well as on sustainable economic progress.
• The Global Dialogue should be informed by the reports of the Independent Scientific Panel.
• It should promote collaboration between initiatives, leverage plurilateral discussions without undermining or duplicating other international, regional and national activity on AI related issues.
• It should provide a space for knowledge-sharing and capacity building for countries and stakeholders with less access to AI capabilities, which can be reflected in summaries of discussions. It should focus on concrete issues and topics.
• The Global Dialogue should therefore take place in the margins of existing multi-stakeholder AI initiatives such as the AI4Good Summit, with provision for virtual formats. It should coordinate with entities such as the OECD and the Global Partnership on AI, and be prepared by the existing coordination mechanism, the UN Interagency Working Group on AI.
• The Global Dialogue should further support in-house dialogues on AI governance within the UN system and create synergies with relevant workstreams. The Global Dialogue should however not lead to the creation of a new institutional structure which overlaps with existing efforts and initiatives.

The EU is fully committed to the implementation of the Global Digital Compact and will engage constructively with all stakeholders in order to establish, based on the principles outlined above, the first Independent International Scientific Panel on AI as well as the Global Dialogue on AI Governance. They should have distinct and separate, but complementary functions. We look forward to a transparent and inclusive process over the next few weeks.

Thank you.