White House Tells OpenAI to Hit Snooze on GPT-5.6 Rollout ⏳

The White House has asked OpenAI to restrict the release of its latest model, GPT-5.6, to government-approved partners only, according to a report from The Information. The company has agreed to the limited initial deployment rather than delay a public launch entirely.

The request follows a series of U.S. actions targeting advanced AI systems over national security concerns. It comes shortly after the administration placed export restrictions on Anthropic, prompting that company to withdraw its Mythos and Fable variants. Those systems had drawn scrutiny in Washington and on Wall Street over their cyber capabilities and associated safety risks. One source told The Information that both the administration and OpenAI's own developers consider GPT-5.6 to be "on par" with Mythos.

In an internal memo to staff on Thursday, OpenAI CEO Sam Altman said the government is giving it "access," though he described the approval framework as a short-term arrangement rather than the company's long-term strategy. The narrower release is being framed internally as a path through current regulatory uncertainty, allowing a public rollout to proceed even as U.S. AI rules remain undefined.

OpenAI has not publicly disclosed the full list of approved partners for GPT-5.6, and the White House has not issued a formal statement detailing the scope of the restrictions. The development underscores how directly federal policy is now shaping product launch timelines for frontier AI developers operating in the United States.