Altman, Amodei and Bengio brief the UN Security Council — and the US says no

The Security Council met as meeting 10228 on Wednesday afternoon under France's September presidency, chaired by French foreign minister Jean-Noël Barrot. It was the Council's first session focused on AI safety risk since 2023, and the briefers were an unusual set: Yoshua Bengio, co-chair of the UN's Independent International Scientific Panel on AI, alongside Sam Altman of OpenAI, Dario Amodei of Anthropic and Clément Delangue of Hugging Face. Three of the four run companies that would be regulated by anything the Council produced.

Each arrived with a proposal. Amodei argued for starting narrow — an agreement not to use AI to help build biological weapons — then building the evaluation and verification machinery that would let states check each other. Altman asked for a shared technical language so countries can compare evidence and verify compliance, incident classification and reporting protocols, and secure government-to-government channels for sharing vulnerabilities. Bengio wanted frontier AI licensed the way medicine, aviation and nuclear power are licensed, with developers proving safety to independent experts before training and before deployment, backed by liability insurance. Delangue, appearing remotely through technical difficulties, asked for stronger monitoring and incident-disclosure standards.

Then Michael Kratsios, director of the White House Office of Science and Technology Policy, told the Council the United States rejects the entire premise. No pause, no new global governance structure; best-practice sharing and American AI exports instead. China is reported to have argued for open-source approaches. Barrot reached for the atomic age as his analogy.

Nothing was adopted. No resolution, no presidential statement, no draft text was tabled — Security Council Report's pre-meeting note lists guiding questions on evaluation and verification, not an instrument. It was a briefing, and the one country whose companies build most of the frontier said it would not be bound.

• Confirmed Security Council meeting 10228, “Artificial intelligence and international security”, convened 23 September under France's presidency and chaired by foreign minister Jean-Noël Barrot; briefers were Bengio, Altman, Amodei and Delangue. UN Web TV
• Confirmed A briefing and debate only — no resolution, presidential statement or other outcome document was tabled or adopted. Security Council Report
• Reported The US position was delivered by Michael Kratsios, who rejected both a pause and any new global governance structure, offering best-practice sharing and AI exports instead. The Next Web