OpenAI's Sam Altman, Anthropic's Dario Amodei and Hugging Face's Clément Delangue on September 23 addressed the United Nations Security Council in New York in the body's first dedicated briefing on artificial-intelligence safety. France's Foreign Minister Jean-Noel Barrot chaired the session.
What Each Speaker Said
Altman argued that any non-zero risk of catastrophic outcomes — "10% or 0.1%" — is unacceptable for frontier AI, and called for international safety standards backed by verification. Amodei asked member states to ban AI-assisted design of biological weapons and warned of misuse pathways that current export-control regimes do not cover. Delangue emphasized that open-source models remain essential for global scientific research and warned against safety regimes that would push development into unregulated jurisdictions.
Where International AI Governance Stands
The Council session followed a summer of parallel AI legislative moves: California's kill-switch executive order, a Chinese State Council white paper on model registration, and today's Sanders-Casar superintelligence bill. No formal Security Council resolution was tabled; France said further sessions are planned before the end of 2026.
Reactions
The Chinese and Russian delegations declined to endorse Altman's call for international incident reporting, arguing that AI governance should sit under existing U.N. specialized agencies. India, Brazil and Ghana supported the creation of a working group on catastrophic-risk thresholds. Council members from Europe echoed Amodei's biosecurity focus.
Reporting based on coverage from Bloomberg, Reuters and industry announcements on September 23, 2026.