On the night of Friday, June 12, Commerce Secretary Howard Lutnick sent a letter to Anthropic chief executive Dario Amodei ordering that Claude Fable 5 and Mythos 5 be made inaccessible to “any foreign national, whether inside or outside the United States.” Because Anthropic couldn’t segregate foreign from domestic users in real time, it pulled both models down for everyone. Eight days later, they remain offline.

The directive is the first time the federal government has invoked export-control authority against a commercial frontier model. That distinction matters more than the immediate disruption. A statute designed for centrifuges and chip lithography has now been pointed at an API.

Anthropic’s public response was studiously narrow. “We believe this is a misunderstanding and are working to restore access as soon as possible,” the company posted on X. The stated justification, relayed verbally rather than in writing, was a “potential narrow, non-universal jailbreak” of Fable 5, which Anthropic had released June 9 as the first of a new “Mythos-class” tier. Mythos 5 itself had been confined to a small group of users under the company’s Project Glasswing cybersecurity program.

The official story strained credulity almost immediately. Katie Moussouris, founder of Luta Security, reviewed the underlying research paper at Anthropic’s request and said in a blog post that the finding “should never have triggered an export control” and that any attempt to patch the described behavior “would only weaken the model for defense.”

A more plausible account surfaced through The Wall Street Journal, as cited by CNBC: the order traces back to conversations between Amazon chief executive Andy Jassy and Treasury Secretary Scott Bessent, in which Jassy reported that Amazon researchers had elicited responses from Fable 5 that could aid in cyberattacks. The jailbreak rationale, in that telling, is the public-facing wrapper on an intra-elite phone call.

The timing is the part that’ll be studied later. Days before the directive, Amodei published an essay calling for “serious and binding regulation” of frontier systems. He got it, in a form that doubles as leverage over a company nearing a $1 trillion valuation and reportedly weighing a public offering. Daniel Remler of the Center for a New American Security, noting the administration’s prior preference for voluntary testing, told CNBC: “This sure looks mandatory if there are going to be consequences for not doing what the government says.”

Anthropic, releasing Fable 5, had warned that “releasing a model this capable comes with risks.” The risk that materialized was regulatory, not technical.

Sources