Sam Altman addresses the UN Security Council on AI risk and international standards

OpenAI CEO Sam Altman addressed the United Nations Security Council on artificial intelligence on September 23, laying out what he called a choice between AI as "a new Renaissance of creativity and discovery" and AI as "a new Industrial Revolution of upheaval and disarray." OpenAI published his remarks as delivered on its own site the same day.

What's new

Altman told the Council that recent progress has made AI's timeline "feel more compressed" and its risks "more immediate," and he said he largely agreed with a warning from Yoshua Bengio, though he narrowed it to two failure modes rather than three. The first is losing control of the future to AI systems that move faster than people can follow or intervene. The second, tied to what he called recursive self-improvement, is that as AI development itself becomes automated, the pace of progress could accelerate beyond what institutions can track.

He committed OpenAI to not accepting outsized technological risk merely because competitive pressure makes slowing down feel costly: "The industry must not accept too much technological risk just because the benefits are too great and that they feel too important to slow down," and said the company has unilaterally slowed development before and would do so again. He also said AI's biggest decisions should not sit with "labs in San Francisco alone" but be shaped through democratic institutions and accountable governments.

On substance, Altman called for four things at the international level: common technical standards for measuring AI capabilities and risks, shared incident-reporting protocols so failures are visible before they compound, secure channels between governments and infrastructure operators to share emerging threats, and safeguards designed to avoid favoring incumbents over new entrants or open models over closed ones.

Context

The speech lands a week after Anthropic CEO Dario Amodei publicly called for the AI industry to deliberately slow down, and after more than 100 companies including OpenAI, Anthropic, and Google signed an open letter on AI cyber defense. It's Altman's most direct engagement yet with a UN body on binding international coordination, rather than the voluntary industry commitments that have characterized most safety pledges to date.

Why it matters

Most AI-safety statements from lab CEOs go to press releases, congressional testimony, or company blogs aimed at regulators in a single jurisdiction. A Security Council address is a different audience: it's aimed at getting governments that routinely disagree with each other to treat AI oversight as a shared-interest problem, the way Altman explicitly framed it. Whether that translates into anything binding is an open question the speech itself concedes, Altman said OpenAI does "not have all the answers," but the ask, common measurement standards, shared incident reporting, cross-border threat sharing, is concrete enough that it gives international bodies something specific to act on rather than a general appeal for caution.