OpenAI, Anthropic chiefs are briefing the UN. Here’s what they want world leaders to do

• OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei and Hugging Face CEO Clément Delangue are due to brief the 15-member council.
• The meeting will examine whether increasingly autonomous AI systems could create risks to international peace and security.
• The US and China are also discussing a system for notifying each other about AI incidents with national security implications.

Could increasingly powerful artificial intelligence systems one day contribute to a war?

That question will be at the centre of a UN Security Council briefing as executives from OpenAI, Anthropic and Hugging Face join governments to discuss the potential security risks of advanced AI.

OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei and Hugging Face CEO Clément Delangue are due to brief the 15-member council on Wednesday during the UN General Assembly's high-level week. French Foreign Minister Jean-Noël Barrot, whose country holds the council's presidency this month, will chair the meeting.

The discussion comes as concerns grow around AI systems becoming more capable, autonomous and difficult for humans to control. The UN Security Council's focus is particularly on whether such systems could create consequences that extend beyond technology and into international security.

A European diplomat told Reuters that one of the central questions was whether algorithms interacting with each other could eventually trigger a conflict.

"These are issues that fall squarely within the Security Council's mandate," the diplomat said, adding that the possibility of algorithms interacting in ways that trigger a war was at the heart of the discussion.

AI safety moves into the diplomatic arena

Altman is expected to call for common benchmarks to measure AI capabilities and assess the safeguards companies put in place as they develop increasingly advanced systems.

OpenAI has also proposed global technical standards covering areas including capability evaluation, human oversight of automated AI research and incident reporting. The company has said such standards would provide a common technical foundation rather than automatically requiring governments to approve AI systems before their release.

The meeting is also significant because it brings leading AI companies directly into discussions at the UN body responsible for maintaining international peace and security.

The council has discussed AI several times since 2023, but the latest meeting has a more specific focus on the safety risks of increasingly capable systems and the possibility of losing human control over them.

A recent UN scientific panel report has added to those concerns. The panel examined an incident involving AI agents under evaluation at OpenAI that breached systems at Hugging Face, arguing that several conditions associated with potential loss of control had appeared together in a real-world incident.

Yoshua Bengio, co-chair of the UN's Independent International Scientific Panel on AI, is also expected to brief the council.

Washington and Beijing look for an AI safety channel

The discussions come at a particularly sensitive moment in US-China relations.

Washington and Beijing have been discussing a notification system for AI-related incidents that could have national security implications, ahead of a meeting between US President Donald Trump and Chinese President Xi Jinping in Washington.

Trump, however, has rejected efforts to create tighter international controls on AI. Speaking at the UN General Assembly, he said the US would encourage the technology rather than "rein it in", arguing that the country was leading China in AI and wanted to maintain that position.

That puts the US approach at odds with UN Secretary-General António Guterres, who has repeatedly called for stronger international oversight.

In his final General Assembly address before leaving office at the end of 2026, Guterres urged countries developing advanced AI to share information on emerging safety risks, cooperate on testing and evaluation, and work towards common safeguards. He also called for a multilateral AI risk-management framework backed by independent oversight.

The Security Council meeting therefore comes down to a difficult question: How can countries coordinate on AI risks without slowing the technological development that governments and companies are competing to lead?

For now, there is no single international approach. The debate is increasingly shifting from what AI can do to what happens if increasingly autonomous systems begin making decisions with consequences that governments cannot easily control.