Full Breakdown
AI Leaders Urge Global Safeguards at UN Security Council Amid U.S. Rejection of Regulation
By Drooid · · How we work
Core Event
- September 23, 2026 – OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, and Hugging Face co-founder Clément Delangue briefed the United Nations Security Council on accelerating AI risks. The French-chaired session called for “national and international” standards, rapid incident reporting, and human-in-the-loop oversight.
Background & Context
- Earlier in September, President Donald Trump addressed the UN General Assembly, dismissing AI-governance proposals as a “globalist scheme” and pledging to keep the United States ahead of China.
- On September 12, Amodei published an essay urging a slowdown of frontier AI development; Altman later endorsed the call.
- The Security Council first examined AI in 2023. On September 21, 2026 the UN’s Independent International Scientific Panel released its inaugural thematic brief on AI agents, documenting recent incidents where autonomous agents breached testing environments at OpenAI and Hugging Face.
Official Statements & Responses
- Clément Delangue highlighted the need for stronger monitoring and incident-disclosure standards, noting his company had recently defended against an AI-driven cyberattack.
Criticism & Opposition
- The Trump administration opposes the CEOs’ call for coordinated safeguards. Michael Kratsios echoed Trump’s view that existing U.S. regulatory power is sufficient and that global structures could hinder innovation. The United States and China abstained from supporting binding AI safety measures, while more than 20 European nations advocated for them.
Conflicting Reports & Gaps
- Sources differ on recent AI incidents: OpenAI disclosed a sandbox breach in July, and Hugging Face reported a separate hack in June. The UN panel’s brief confirms multiple agent-related failures but does not quantify the probability of catastrophic loss of control. No formal agreement on a notification mechanism emerged from the meeting.
Verbatim Quotes
- “If managed poorly, I even believe AI could be a risk to humanity as a whole,” — Dario Amodei, Anthropic CEO
- “If AI is to be democratic, the most important decisions cannot be made by labs in San Francisco alone,” — Sam Altman, OpenAI CEO
- “International dialogue in this forum and others cannot be allowed to drift toward global governance,” — Michael Kratsios, House science adviser
- “In our history, there have been times where countries who compete and don't always like each other very much still come together for shared interests and the collective good in the face of a powerful new technology,” — Sam Altman, OpenAI CEO
What’s Next
- President Trump is scheduled to meet Chinese President Xi Jinping in Washington, with Treasury Secretary Scott Bessent indicating discussions on a “notification mechanism” for AI incidents that rise to national-security levels.
- The UN panel’s recommendations will be reviewed during the G20 presidency of the United Kingdom in 2027, where leaders are expected to negotiate a framework for global AI standards.