Key Highlights

  • The chief executives of OpenAI, Anthropic and Hugging Face addressed the UN Security Council on the risk that advanced AI systems could eventually escape meaningful human oversight.
  • Sam Altman intended to push member states toward adopting shared benchmarks for measuring both AI capability and the safeguards companies have in place.
  • The US and China have separately begun discussing a notification arrangement for AI incidents that cross into national-security territory, ahead of a Trump-Xi meeting in Washington.
  • Trump used his own UN address a day earlier to reject any "globalist scheme" to regulate AI, saying his administration would rather encourage the technology's growth than slow it down.

Industry's Top Voices Make Their Case to the World's Security Body

The Security Council spent Wednesday hearing directly from the people building the most capable AI systems in existence, as concern grows that such systems could eventually begin improving themselves faster than anyone can supervise. OpenAI's Sam Altman, Anthropic's Dario Amodei and Hugging Face co-founder Clement Delangue all addressed the 15-member body, with France's foreign minister Jean-Noel Barrot presiding over the session during the UN's annual high-level gathering of world leaders in New York.

Altman came prepared to press a specific ask: that governments settle on common benchmarks for evaluating how capable a given AI system actually is, and for judging whether the safeguards a company has built around it are adequate, according to an OpenAI representative who briefed reporters beforehand. Council members with the deepest stakes in the outcome, the US and China chief among them, were also expected to weigh in during the session.

The Real Fear: Machines Talking to Machines

A European diplomat who wasn't authorized to speak on the record framed the session's purpose plainly: surface the concrete ways AI could spiral out of anyone's control in a manner that actually threatens national or international security, rather than treating the topic abstractly. "These are issues that fall squarely within the Security Council's mandate," the diplomat said, adding that the real question underlying everything else was whether interacting AI systems could someday set off a conflict without a human ever making that call.

That concern isn't purely theoretical to the two governments best positioned to shape how it plays out. Ahead of Thursday's meeting between Trump and Chinese leader Xi Jinping in Washington, officials from both countries have begun sketching out a system for flagging AI-related incidents that rise to the level of a national security concern, an early, narrow area of cooperation running underneath a far larger rivalry.

A Head-On Collision With Washington's Posture

The session landed awkwardly against what Trump told the same assembly just a day before. He said the US "rejects any attempt to construct a globalist scheme to control" artificial intelligence, insisting the country's lead over China and everyone else was too valuable to jeopardize. "I'm not going to stifle growth of something that will be bigger than the Industrial Revolution," he said, adding that any oversight would run through the Department of Justice rather than a broader regulatory push. "We're going to encourage it, not rein it in."

That stance sits in direct tension with UN Secretary-General Antonio Guterres, who has spent years pushing for coordinated global oversight of AI and warning its risks can't simply be waved away. Guterres assembled a 39-member advisory panel back in 2023 drawing from the US, Russia, Japan and elsewhere, and the broader General Assembly followed up last year by standing up a permanent Independent International Scientific Panel. That panel's co-chair, Canadian researcher Yoshua Bengio, often counted among AI's founding figures, also briefed the council Wednesday. It wasn't the body's first pass at the subject: China warned in 2023 against letting the technology become a "runaway horse," the US countered with concerns about AI-enabled censorship, and then-Secretary of State Antony Blinken chaired a follow-up session in 2024.