OpenAI and Anthropic bosses push UN for global terms on AI

The biggest AI names in the room are no longer just talking about products and speed. At a UN meeting this week, the leaders of OpenAI, Anthropic and Hugging Face argued that the current pace of AI development has gone far enough to demand international coordination.

Sam Altman wanted common standards for judging what AI systems can do, what risks they carry, what safeguards they include, and how much human oversight stays in place. He also pushed for faster incident reporting so failures can be tracked before they turn into something worse. His argument was blunt: if AI is meant to serve everyone, the key decisions cannot be left to labs in San Francisco alone.

Dario Amodei said AI could become a risk to humanity if it is handled badly, and called for even tougher safety standards as the technology advances. Clement Delangue made a similar case for stronger monitoring and disclosure rules. The message from all three was the same: this is not a problem one company can manage by itself.

But the US government’s line at the same meeting ran in the opposite direction. Michael Kratsios, a technology adviser to Donald Trump, said the speed of AI development is indeed rising and the risks are real, but that was not reason enough to pause development or build new global governance structures. Trump has also made clear he wants the US to keep its lead over China in AI, and has rejected the idea of a slowdown.

That split is the real story here. The companies building the systems are asking for shared rules before the mess gets bigger. Washington, at least under Trump, is choosing competition first and coordination later. And that usually means the rest of the world gets to deal with the consequences after the fact.