White House to Release AI Model Release Standards Next Week, Rolling Back Trump's Earlier Light-Touch Promise

According to the Financial Times, the White House plans to release a voluntary standard for advanced AI model releases as early as next week, aimed at setting benchmarks for models with top-tier cybersecurity capabilities and establishing timelines for future releases. The U.S. AI Standards and Innovation Center and the National Security Administration (NSA) will play key roles in setting and overseeing the standard.

Recent actions by the Trump administration have directly intervened in AI labs: Anthropic's latest model faced export controls on June 12 over misuse and cybersecurity concerns, which were lifted only this week; OpenAI was required to release GPT-5.6 only to government-selected groups, with broader release expected as early as next week; Google is also in talks with the government before releasing a new coding model with stronger capabilities.