The chief executive of OpenAI urged the adoption of unified safety benchmarks for artificial intelligence during a meeting with the United Nations Security Council on Wednesday. Sam Altman argued that nations must establish common global AI safety standards to manage risk, but cautioned that governments should avoid implementing regulations that would inhibit technical innovation.

OpenAI does not support the creation of a specific international regulator, favoring democratic oversight and basic safety requirements instead. Anthropic CEO Dario Amodei joined the session via video, while representatives from DeepSeek briefed the council on international security and the risk of advanced models slipping beyond human control. Moonshot also received an invitation to speak at the briefing.