UN Security Council to hear from OpenAI, Anthropic and DeepSeek on AI risks
The United Nations Security Council will be briefed on artificial intelligence risks on Wednesday, Sept. 23, 2026, by some of the executives whose companies build the technology — including OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei. France convened the session and holds the council’s rotating presidency this month, according to Decrypt.
Foreign Minister Jean-Noël Barrot will chair the 15-member body’s meeting, which Decrypt reported is built around a concept note centered on AI systems slipping beyond human control. The session takes place two days before President Donald Trump sits down with Chinese President Xi Jinping in Washington on Sept. 24, with AI safety reportedly on that summit’s agenda alongside trade. Cointelegraph reported that the UN confirmed Altman and Amodei’s participation, citing CNBC.
Also read: Hana Bank issues $100M digital bond on Euroclear D-FMI
Key facts
- Altman, Amodei, Hugging Face CEO Clément Delangue and AI safety researcher Yoshua Bengio are all scheduled to brief the council; Amodei will join remotely rather than in person, according to Decrypt.
- Chinese AI firms DeepSeek and Moonshot have been invited to make statements, per Reuters via Cointelegraph, and DeepSeek is expected to take part — though founder Liang Wenfeng does not plan to attend, the report said.
- Barrot will chair the session, which France organized while holding the council’s rotating presidency; the meeting follows Amodei’s Sept. 12 essay, “We Must Pace the Frontier,” which urged the industry to slow the build-out of more capable frontier models and said Anthropic would give outside evaluators permanent, staff-level access to check its safety promises.
- The Security Council first formally discussed artificial intelligence in 2023, Reuters reported, and Wednesday’s session is not expected to produce a binding deal.
- Days before the briefing, King Charles hosted executives from Anthropic, OpenAI, Nvidia and Google DeepMind at his Scottish estate for an AI safety summit that produced conversation but no publicly known agreements.
A council built for wars hears from tech CEOs
The briefing follows a year in which Amodei has called China’s government the industry’s biggest threat, per Decrypt. He described China as “the autocratic country with by far the most advanced AI capabilities” and argued Washington should keep restricting exports of advanced chips to Beijing even as it asks Beijing to cooperate on global safety rules. China’s Foreign Ministry called the essay counterproductive, and state media branded it a “Cold War playbook” aimed at containing China.
The invitation structure itself is the story’s main tension: the same session seats DeepSeek, the Chinese startup that wiped roughly $600 billion off Nvidia’s market cap when it launched a cut-price model in January 2025, and Moonshot, the developer of the Kimi models, alongside the American labs. Cointelegraph reported that the meeting will examine growing concerns over AI’s impact on international security — concerns that also cover systems that could run cyberattacks, sway elections or design weapons without a person in the loop.
Also read: Barry Silbert Backs Bittensor as AI Successor to Zcash
Decrypt reported that Altman and Elon Musk both said they agreed with Amodei’s essay within hours of its publication, while Trump rejected the slowdown push, calling AI-takeover warnings a “hoax” tied to the “Radical Left” and accusing Amodei of wrongdoing without offering evidence. That split — parts of the industry asking to slow down while the White House resists — is the backdrop the council will hear from on Wednesday. The reports do not uniformly describe Amodei’s attendance: Decrypt states he will join remotely, while Cointelegraph described him only as expected to take part.
Why it matters
The council holds the power to authorize sanctions and military action, so its agenda signals which technology risks governments now treat as security matters rather than regulatory ones. For AI labs, the session puts safety commitments made in essays and blog posts in front of the body that could eventually write the rules they operate under. Compared with the council’s first AI discussion in 2023, Wednesday’s lineup shifts the conversation from governments talking about the technology to the companies building it talking to governments, with Chinese and American labs in the same room days before their leaders meet.
What to watch
Watch whether the Sept. 24 Trump-Xi summit in Washington produces any joint language on AI safety, and whether the council takes up a follow-on process after Wednesday’s statements. Neither report describes a vote or a draft resolution tied to the session.
Sources: Decrypt, Cointelegraph
