The U.S. government has partially lifted its restrictions on Anthropic, clearing the company to release its Mythos AI model to a limited set of recipients, according to reporting first published by Semafor and picked up by Reuters, CNBC and others.
Semafor, in an exclusive, reported that the powerful Anthropic model is being released to some U.S. companies. CNBC reported that the Trump administration is allowing the release to a mix of private companies and government agencies. According to 9to5Mac, Anthropic has been cleared to distribute Claude Mythos 5 to more than 100 U.S. institutions, while Ynetnews framed the move as access for "trusted partners."
The decision is a partial reversal rather than a full reopening. Neowin reported that while Washington is permitting Mythos to go out, it is keeping a separate model, Fable 5, off the market. DW.com and The Express Tribune similarly described the action as a "partial" or "limited" release, with The Express Tribune referring to the model as Mythos 5.
The sources here are headlines and summaries rather than full filings, so the precise terms — which companies and agencies qualify, what conditions apply, and why Fable 5 remains restricted — are not spelled out in the material available. What is consistent across the reports is the shape of the move: a controlled, partner-by-partner rollout of one model while another stays blocked.
Why it matters: governments are increasingly treating advanced AI models like strategic technology, deciding case by case who may use the most capable systems — and this selective clearance shows that access to frontier AI is now something officials grant and withhold, not a product simply sold to anyone.