Trump Administration Restricts OpenAI's GPT-5.6 Sol and Anthropic's AI Models to Approved Customers Amid Cybersecurity Review

The Trump administration has inserted itself into AI product releases, requiring OpenAI and Anthropic to limit access to their newest models to government-approved customers. OpenAI's GPT-5.6 Sol — described as better at finding vulnerabilities than exploiting them — is currently available to roughly 20 undisclosed customers blessed by Washington. Anthropic, meanwhile, had its Mythos 5 model partially reinstated after a two-week ban, while its reportedly safer sibling, Fable 5, remains offline with no clear path back.

The legal basis for all this is shaky: Trump's executive order on AI oversight describes developer participation as "voluntary" and the review framework hasn't been fully developed. Stanford cybersecurity expert Alex Stamos was blunt, saying "pretty much nobody in the cybersecurity industry believes that there's any factual basis for this action" and calling the restrictions "about the dumbest thing" the administration could do if it actually wants to beat China in AI. Rep. Lori Trahan put it more procedurally: "No law. No process. No oversight. Just appointees in Washington deciding who's in and who's out."