Claude Fable 5 and Mythos 5 entered their eighth day offline on Friday, with Anthropic and the Commerce Department still negotiating the terms under which the company’s two newest models can return to paying customers. The directive landed at 5:21 p.m. ET on June 12, instructing Anthropic to halt access for “any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.” Unable to segregate users that quickly, the company pulled both models for everyone.

Per Bloomberg, citing an unnamed U.S. official, this is the first time the Trump administration has used export-control authority to force a leading AI lab to withdraw a deployed model from commercial customers. The mechanism is the news. Export controls have, until now, governed chips and weights moving outward; here they’ve been pointed at an inference endpoint serving global enterprise traffic.

The government’s verbal justification cited a jailbreak against Fable 5’s safeguards. Anthropic, in its statement, called the finding “a small number of previously known, minor vulnerabilities” surfacing in other publicly available models including OpenAI’s GPT-5.5, and said bluntly: “We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.”

The current standoff sits on top of an older one. In March the Pentagon designated Anthropic a “supply chain risk,” a label the company is already contesting in federal court. Claude Opus 4.8 remains available.

The financial backdrop sharpens the stakes. Fortune reports Anthropic was valued at $965 billion in a recent round and filed confidentially for a public listing earlier this month. An indefinite outage of two flagship models is now a disclosed pre-IPO risk, negotiated bilaterally with the executive branch. Asked about the talks at the G7, President Trump said they were “going fine.”

Sources