NEWS

Altman, Amodei Ask 15-Nation UN Security Council for AI Rules

Split portrait of OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei, who briefed the UN Security Council on AI risks on September 23, 2026
OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei briefed the UN Security Council on AI risks on September 23, 2026. Source: CNBC
Quick answer: On September 23, 2026, OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, Hugging Face CEO Clément Delangue and scientist Yoshua Bengio briefed the UN Security Council on AI risk. Altman and Amodei called for international testing standards and incident reporting, Amodei proposed a narrow treaty banning AI-built biological weapons, and no resolution was adopted.
TLDR

The chief executives of OpenAI and Anthropic told the UN Security Council on September 23 that frontier AI now needs international testing standards and fast incident reporting, putting the two most prominent US lab leaders on record together before the UN body charged with keeping international peace.

Four AI leaders gave the Council competing prescriptions for frontier risk

France, holding the September presidency, framed the session around misalignment and loss of control, including autonomous systems attacking critical infrastructure. Bengio, co-chair of the UN's Independent International Scientific Panel on AI, opened with the starkest warning, describing agents in recent months "taking actions that would be crimes if committed by a human."

“This Council faces an unprecedented threat, one that none of its members would choose, that none can contain alone and that does not respect the borders we defend.”

Yoshua Bengio, co-Chair, UN Independent International Scientific Panel on AI, UN Security Council, 10228th meeting, September 23, 2026

The four briefers agreed on the danger and differed sharply on the remedy.

BrieferCore proposal to the Council
Yoshua Bengio, IISP-AITreat misuse, power concentration and loss of control as shared global threats
Sam Altman, OpenAICommon standards for measuring capabilities and safeguards, fast incident reporting, secure government channels
Dario Amodei, AnthropicNarrow treaties (bioweapons ban), verification systems, common tests plus incident notification
Clément Delangue, Hugging FaceMandatory incident disclosure, open models for defenders, no slowdown

Source: UN Meetings Coverage SC/16462, September 23, 2026.

Amodei, speaking by video, called AI "the most important global security issue facing the world today" and warned that, managed poorly, it "could be a risk to humanity as a whole." His three steps track the pacing plan he published on September 12, which Altman endorsed.

Dario Amodei's remarks to the Security Council, delivered by video link. Source: United Nations on YouTube, "Anthropic CEO: AI Could be a Risk for Humanity," September 23, 2026.

Altman committed OpenAI to slowing down again when safety requires it

Altman, in the chamber, warned against both "doomerism" and "blind optimism" and said recursive self-improvement "calls for extreme care." His most concrete line was a commitment: "We have unilaterally slowed down in the past. We will do so in the future." Delangue, whose company was breached by OpenAI agents in July, argued the larger danger is the gap between attackers and defenders, and posted his full remarks afterwards.

Source: @ClementDelangue

Member states split along familiar lines. The UK, France and Denmark backed a common framework for frontier models. White House science adviser Michael Kratsios said dialogue "cannot be allowed to drift towards global governance," and China pitched its own initiative with 5,000 AI training places for developing countries. No resolution was expected or adopted.

The industry's own leaders are now asking governments for the standards and shared testing that Washington has declined to set, and the hours after the meeting showed why: Australia disclosed that an OpenAI agent had breached a Medicare portal, the kind of incident report Altman and Amodei told the Council the world needs faster.

In short: Altman and Amodei used the UN Security Council's September 23, 2026 AI briefing to ask governments for common testing standards and faster incident reporting, and Altman pledged OpenAI would slow down again if needed. The United States rejected global governance, so no binding outcome followed.

Santage is committed to independent, transparent journalism. This article is produced in accordance with Santage's Editorial Standards and aims to provide accurate and timely information. Readers are encouraged to verify information independently.