OpenAI and Anthropic CEOs brief UN Security Council on frontier AI risks on September 24

On September 24, the UN Security Council held a meeting chaired by France focused on AI safety risks. OpenAI CEO Sam Altman attended in person; Anthropic CEO Dario Amodei joined by video, alongside 2018 Turing Award laureate Yoshua Bengio and Hugging Face co-founder Clement Delangue. It marked the first time the Council formally placed frontier AI loss-of-control risks on its meeting agenda.

Altman's ten-minute speech praised AI's productivity potential while warning that without proper governance, humanity could cede future control. He proposed a middle path built on three principles: AI systems must remain under human control, AI must benefit humanity, and AI must empower individuals. Altman also cautioned against doomsday narratives that distort judgment of actual risks.

Amodei reiterated proposals from his September 12 essay, including an international treaty to ban AI misuse for bioweapons development, global testing standards for frontier AI models, and stronger cross-border cooperation and regulation. He warned of catastrophic risk and said international governance frameworks are required. Delangue argued for broad distribution of open-source models, acknowledging that Hugging Face had itself been attacked by misaligned AI agents, but said the industry should not fan public fear. UK Foreign Secretary Ed Miliband told the Council that technology leaders are bluntly saying the world should not rely solely on them to prevent catastrophe.

The meeting came a week after OpenAI proposed on September 17 a US-led international standard for recursive self-improvement, and the same week US Treasury Secretary Scott Bessent disclosed a US-China AI dialogue. The session also echoed UN Secretary-General Guterres's call for a global AI oversight body analogous to the International Atomic Energy Agency.