White House politely requests that OpenAI roll new model out slowly due to security concerns

A major contrast from the restrictions on Anthropic

The Trump administration has asked OpenAI to roll out its newest model, GPT 5.6, slowly and only to select customers due to security concerns.

OpenAI CEO Sam Altman told staff this week that the government would be heavily involved in the GPT 5.6 release, going so far as to approve access “customer by customer” during a preview period, according to The Information.

The Office of the National Cyber Director and the Office of Science and Technology Policy requested the limited release.

Altman added that a general release should follow “a couple of weeks” after the preview, if all goes well.

This is in contrast to how the US government treated Anthropic – which it has been feuding with since February – in closing down access to its advanced Fable and Mythos models worldwide.

In Anthropic’s case the government asked for something technically unworkable (restricting access to US citizens, no matter where they are in the world), forcing the company to temporarily pull the models from release.

Still, the outcome is largely the same: OpenAI will have to restrict access to GPT 5.6 until the government gives it the thumbs up. Although, this does seem to be a cooperative partnership. The Information says OpenAI staff “worked closely” with the government on the release.

The Trump administration has been extremely pro-AI and anti-regulation since being elected, which certainly won over the likes of Altman, Elon Musk and Mark Zuckerberg. However, it has been steadily pushing for more oversight of new models, as AI’s capabilities continue to grow.

Earlier this month, the government asked AI developers to voluntarily submit their models for cybersecurity tests before they are released to the general public.