OpenAI and Anthropic CEOs Brief the UN Security Council on AI Risk
Sam Altman and Dario Amodei addressed the UN Security Council on artificial intelligence and international security on 23 September 2026.
The session
The session was the Council's 10228th meeting. Its full title was "Maintenance of international peace and security: Artificial intelligence and international security," and it was listed as a Briefing. It was scheduled for 3:00 PM that day in the Security Council Chamber, according to the Council's programme of work. UN Security Council
France held the Council presidency in September 2026. The presidency scheduled the 10227th meeting, on maintenance of peace and security of Ukraine, also as a Briefing, for 11:00 AM on 23 September 2026 in the same chamber. A resumed 10227th meeting on Ukraine was scheduled to follow the adjournment of the 10228th meeting.
The speakers
Altman is Chief Executive Officer of OpenAI. Amodei is Chief Executive Officer of Anthropic. Their joint appearance on 23 September brought the leaders of two commercial AI laboratories directly into the Council chamber. Reuters
Altman told the Council that AI systems must remain under human control. He called for a "middle path" on AI, starting with the principle that "people must remain at the center of AI decision making." OpenAI later published his remarks on its website, describing them as addressing AI safety, human control, and international cooperation. OpenAI
Amodei told the Council that AI could be a risk to all humanity. That warning was carried in same-day reporting of the briefing, which framed the session in stark terms about control and scale. PBS
How people could watch
The United Nations promoted a "Watch live now on UN Web TV" link for the AI briefing. The Council website also listed a news story dated 22 Sep 2026 titled "LIVE: OpenAI and Anthropic brief Security Council amid real and imminent threat posed by runaway AI."
The broader context here is format and placement. A briefing imposes no decision. It creates a record, puts expert testimony before all 15 members, and lets a presidency set a theme without forcing negotiation on language, a bit like calling a study session rather than a test. Holding the AI briefing between two Ukraine sessions kept the Council's standing conflict work moving while opening the afternoon for a cross-cutting risk.
Looking at what this means for diplomacy, the notable element is who briefed. The Council more often hears from UN officials, envoys, or humanitarian chiefs. Hearing directly from Altman and Amodei gives member states unmediated access to builders of frontier systems, but it also concentrates technical authority in private voices. The questions to weigh are familiar ones. Which states will seek follow-up products, whether a presidential statement or further briefings, and which will prefer to keep AI governance in other bodies. Whether calls for human control and international cooperation converge into Council language. And how the principle that people remain at the center of decision making translates, if at all, into expectations about deployment, oversight, and accountability.
