White House Wants Early OpenAI Rollout Screened by U.S. Agencies

The Trump administration has asked OpenAI to slow the rollout of its most advanced AI model so companies getting early access can first be screened by the U.S. government. According to the Financial Times, OpenAI plans to give about 20 companies initial access to GPT 5.6, and each is expected to undergo a government background check.

People familiar with the matter said the phased release request came from the Treasury Department, Commerce Department, and White House offices focused on cybersecurity and science and technology policy. The concern is that newer AI systems could be used to break existing cyber defenses, disrupt critical industries, or transfer advanced technology to countries such as China.

The request is part of the administration's effort to limit AI-related risks while preserving U.S. leadership in the field. The White House has already unveiled plans for a framework meant to ensure new models are safe, though that framework is not binding.

President Donald Trump has repeatedly said he opposes heavy regulation of AI companies, arguing they should be "free to innovate and invent." The U.S. request comes as the Commerce Department this month imposed export restrictions on Anthropic, OpenAI's rival, barring foreign companies from using its new models, Fable and Mythos, over security concerns. Separately, the New York Times reported that OpenAI is planning to delay its planned IPO until 2027, amid sharp recent declines in the share price of SpaceX, controlled by Elon Musk, which completed the largest IPO in history two weeks ago.