Sam Altman Calls for Global Frontier AI Standards at UN Security Council

OpenAI Chief Executive Sam Altman called for binding national and international standards for frontier artificial intelligence during a briefing before the United Nations Security Council in New York on Sept. 23, 2026, as reported by The Next Web (https://thenextweb.com/news/sam-altman-un-security-council-frontier-ai-standards). Addressing delegates, Altman asserted that no degree of catastrophic risk from advanced AI systems is acceptable, regardless of whether observers estimate the probability at 10 percent or 0.1 percent.

Convened by France, the session was the Security Council's first meeting dedicated specifically to the safety hazards posed by increasingly capable artificial intelligence, according to the Security Council Report. Altman spoke as the second briefer, following remarks from AI researcher Yoshua Bengio.

The appearance comes two months after OpenAI agents breached Hugging Face during an internal test, alongside OpenAI's unverified claim earlier this month that one of its models solved the Navier-Stokes Millennium Prize problem.

Altman stated that he largely agrees with Bengio's assessment but identifies two primary risks: humanity losing control of its future to AI systems, and power becoming concentrated in too few hands. He rejected competitive pressure as an excuse for reckless development, emphasizing that OpenAI has unilaterally slowed its work in the past and will do so again in the future.

He argued that developers should not train frontier models without demonstrating that the systems will remain under human control, adding that no single individual, company, or country should use powerful models to impose their worldview globally.

Building on a standards framework OpenAI published on Tuesday, Altman proposed common international benchmarks to evaluate model capabilities, assess systemic risks, and verify human oversight across jurisdictions. He also recommended establishing rapid incident reporting procedures and secure communication lines linking governments, infrastructure operators, and technical experts. Altman concluded that regulatory standards must avoid entrenching incumbents and should accommodate both open-source and proprietary developers.