OpenAI Presses Safety Standards Before Altman's UN Briefing

Sam Altman is scheduled to brief the UN Security Council on Wednesday, September 23, at an open session on artificial intelligence and international security convened during the UN General Assembly's high-level week in New York. His remarks are expected to concentrate on OpenAI's safety steps and the case for shared global standards, per reports previewing the appearance.

The trip caps a run of coordinated messaging from OpenAI on where the standards line should sit. On September 9, chief global affairs officer Chris Lehane reversed the company's prior line against binding US federal rules, endorsing common testing, independent model assessment, cybersecurity requirements and mandatory reporting of serious safety incidents. "The prospect of AI-accelerated AI development demands more than voluntary commitments," Lehane wrote. The proposal would apply only to companies developing the most powerful models and would not restrict open-source systems.

Chief scientist Jakub Pachocki had set the technical stakes days earlier, in an essay titled "An Alien Mind." "Currently, I believe that no lab has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer," he wrote. OpenAI has also said this month it does not yet know how to "safely get all the way to aligned, full RSI" and cannot assume that progress in alignment will keep pace with capability gains.

The push does not land in a clean room. On September 11, 25 Fields Medalists, recipients of mathematics' top prize, published a joint declaration at mathandai.org titled "A Severe Misalignment of AI in Mathematics," arguing that AI labs are treating famous open problems as capability benchmarks. The immediate catalyst was OpenAI's September 8 claim that an internal model coordinating roughly 10,000 agents produced a Navier-Stokes-related proof over 88 hours.