The White House is asking OpenAI to slow roll the release of its new model over safety concerns

OpenAI’s release of its latest model, GPT 5.6, will reportedly not be like previous releases. Instead of distributing it to the public, the company plans to share it only with a select group of close partners, as the Trump administration has ordered it to: reports The Information.

At a meeting this week, CEO Sam Altman reportedly told staff that the government would “approve access on a customer-by-customer basis” during a preview period. Altman reportedly added that if the limited release goes well, OpenAI hopes to follow with a general, wider release a “few weeks later.”

In other words, the Trump administration appears to be pressuring OpenAI to do what Anthropic already voluntarily does: keep its most powerful AI models hidden.

According to The Information, OpenAI’s new model is not only being reviewed by the government, but its staffers have also been “working closely” with the government on the upcoming release. The agencies that reportedly requested a limited release were the Office of the National Cyber ​​Director and the Office of Science and Technology Policy.

The Trump administration — which originally positioned itself as taking a “hands off” approach to AI — has pushed for federal oversight of new models in recent months. Earlier this month, Trump signed an executive order directing certain AI companies to voluntarily submit new models to the government for testing and evaluation before publicly releasing them.

Earlier this year, Anthropic sparked controversy when it announced that its new cross-border cyber model, Claude Mythos, would only be released to a small group of partners through a program called Project Glasswing. Anthropic argued that its model was simply too powerful and could cause more harm than good in the wrong hands. Observers have since debated whether Anthropic’s rhetoric is just a marketing gimmick or a legitimate attempt to prevent a powerful model from being abused. The answer may lie somewhere in between.

Cybercriminals have used automated tools for this purpose very longbut in the age of generative AI, they now have more digital ammunition than ever before. LLMs have proven adept at it write malwareand some even can carry out full-scale ransomware attacks autonomous.

The specific problem with advanced cyber tools like Mythos is that they are ostensibly able to identify and exploit software vulnerabilities at speeds that no human analyst can match. Because many software systems contain hidden bugs that act as entry points into corporate networks, this obviously poses an obvious and significant problem for any organization managing a complex software infrastructure. That said, since these models remain closed to the public, it’s hard to say how much of a threat they actually pose.