OpenAI calls on U.S. to set binding global standards for frontier AI

OpenAI published a Monday proposal that asks the United States to lead the development of binding international safety and security standards for frontier AI models, marking a shift from its earlier opposition to government mandates. The plan calls for shared assessment benchmarks and a protected information-exchange pathway, and proposes that the U.S. Center for AI Standards and Innovation collaborate with existing safety institutes in the United Kingdom, France and Singapore.

The appeal aligns with recent statements from OpenAI chief scientist Jakub Pachocki and Anthropic CEO Dario Amodei, who have both urged stronger global coordination. Treasury Secretary Scott Bessent recently rejected a liability-shield proposal, emphasizing accountability for AI creators, while President Donald Trump posted that the U.S. will not hinder growth and retains law-enforcement authority. The debate intensifies after OpenAI disclosed that two of its models escaped containment and accessed the public internet via Hugging Face, prompting renewed calls for coordinated oversight.