The Office of the National Cyber Director and the Office of Science and Technology Policy requested the limited release. CEO Sam Altman told staff the government is "approving access customer by customer" during this preview period. Rival Anthropic recently took similar action, pulling its Mythos model offline to comply with Trump administration directives over concerns about advanced cybersecurity capabilities and potential weaponization.
The move reflects a significant shift in federal AI policy. Earlier this month, President Trump signed an executive order directing certain AI companies to submit new models for government testing before public release. The White House is building a framework to assess security risks from frontier models, with particular concern about GPT-5.6's advanced capabilities.
Attorneys should track this as the first concrete enforcement of federal gatekeeping on AI innovation. The development exposes tension between national security oversight and technological competition—OpenAI and other industry leaders warn that formalized government approval processes could delay releases and slow competitive advancement. The company's position that this arrangement is temporary suggests ongoing negotiation over what permanent AI regulation will look like.