An OpenAI government review of its most powerful AI models is now on the table. OpenAI says it will back President Trump’s new AI executive order, which asks developers to give federal agencies an early look at frontier models before release.
George Osborne, OpenAI’s head of countries, confirmed the decision. He told CNBC that “it’s quite right that democratic governments have a big role to play in how this technology is used.”
What the OpenAI government review covers
The OpenAI government review sits inside a broader policy. The order, signed June 2, is titled Promoting Advanced Artificial Intelligence Innovation and Security. It asks developers to give the government access to so-called covered frontier models.
That access can run for up to 30 days before a model reaches other partners. During the window, agencies run a classified benchmarking process. They probe a model’s advanced cyber capabilities, and the NSA director decides whether it counts as a covered frontier model. That label could limit how a model is distributed or sold.
Voluntary, and scaled back
The framework is voluntary by design. The order rules out any mandatory license, preclearance or permit to build or release a model. So OpenAI is choosing to opt in, rather than being forced.
The final version is also softer than earlier drafts. Its review window shrank from 90 days to 30 after heavy tech-industry lobbying, and some critics call the result underwhelming. Still, getting a top lab to sign on hands the White House an early endorsement.
The timing fits OpenAI’s wider strategy, too. The company has pushed for clear federal rules instead of a patchwork of state laws, so backing a national order plays to that goal. It also lets OpenAI shape how oversight works while the framework is still young.
The next question is whether rivals follow OpenAI’s lead. Related: OpenAI is also making ChatGPT’s memory more personal.
