The United States government is deep in negotiations with major artificial intelligence companies to develop voluntary guidelines governing how new AI models are released to the public, the Financial Times reported Wednesday, citing unnamed sources familiar with the discussions. An official announcement could come as soon as next week.
Federal officials have been stepping up oversight of new AI model releases amid concerns that cutting-edge artificial intelligence technology could be exploited by military or intelligence agencies in China, Russia, or other nations considered adversarial to the U.S.
According to the Financial Times report, the proposed standards would establish performance benchmarks for advanced AI models along with release timelines, and would clarify who is permitted to access those models — both within the United States and internationally.
Reuters, which first covered the story, was unable to independently confirm the report. The White House, Anthropic, and OpenAI did not respond to requests for comment made outside of normal business hours.
The negotiations follow an executive order signed by President Donald Trump in June, which directed federal agencies to collaborate with leading AI developers on testing advanced models prior to their public release and drafting formal standards for the industry.
Google has separately been in talks with the government ahead of plans to release advanced coding models with significantly enhanced capabilities, a source told Reuters Wednesday. That source added that Google is also participating in the broader industry-wide standards discussions. The Financial Times was first to report those details.
The U.S. Commerce Department also made headlines this week when it lifted export restrictions Tuesday on Anthropic’s most advanced AI models — known as Fable and Mythos — less than three weeks after suspending their export over national security concerns.
OpenAI has faced its own set of restrictions. The company delayed a full public rollout of its GPT-5.6 model last week at the request of the U.S. government, restricting access to a limited group of pre-approved partners. Both OpenAI and Anthropic are currently preparing for initial public offerings.
