UN Security Council AI Briefing: Rival Labs Testify
3 min readThe UN Security Council opened its doors on Wednesday to the people actually building frontier AI systems, and for the first time that group included rivals from both sides of the US and China divide. France, which holds the Council presidency for September, convened the 15 member body for a session on artificial intelligence and international security during the General Assembly’s high level week in New York.
Who Is in the Room
OpenAI chief executive Sam Altman is briefing the Council alongside Anthropic CEO Dario Amodei, Hugging Face CEO Clement Delangue, and Yoshua Bengio, the University of Montreal professor who co-chairs the UN AI Advisory Body. The Chinese labs DeepSeek and Moonshot AI were also invited to make statements, according to Quartz. DeepSeek founder Liang Wenfeng is not expected to appear in person, though arrangements could still shift.
Why This Session Is Different
The Council has taken up AI before, but those earlier sessions were built around diplomats, academics and UN officials talking about the technology. Seating Chinese and American frontier developers in the same chamber to address the same body is new. It is also a tacit admission of something governments have been slow to say plainly: the pace of capability advance is set inside a handful of private labs, not inside any treaty.
The Slowdown Fight Follows Them to New York
The timing is loaded. On September 12, Amodei published an essay calling for industry wide cooperation to decelerate capability gains in favor of safety work, and Altman, Elon Musk and Demis Hassabis all signaled agreement the same day. A week later, four paying subscribers filed a federal antitrust class action in the Northern District of California arguing that competitors cannot lawfully decide among themselves that competition is too dangerous. Chinese officials have separately dismissed the slowdown push as a containment tactic. All three positions now arrive at the same table.
Why It Matters
A Security Council session produces no binding rules by itself, and any resolution would still face the veto. What it does produce is a shared record. Once the labs have stated their own risk assessments on the Council floor, those statements become something diplomats can cite, compare and hold them to later. Watch for whether the Chinese labs’ framing of catastrophic risk lines up with their American counterparts, because that overlap is the only foundation a future agreement could be built on.
Today’s meeting will not settle who governs frontier AI. It does mark the moment the question stopped being theoretical for the world’s top security body.