At 5:21 p.m. Eastern on June 12, Anthropic PBC received a Commerce Department export-control directive ordering it to bar all foreign nationals from Fable 5 and Mythos 5. By the next morning, both models were disabled worldwide. Claude Opus 4.8 remained online; the company’s two most capable systems didn’t.

The rationale, as relayed verbally by officials and later recounted in Anthropic’s own statement, concerned a “potential narrow, non-universal jailbreak” of Fable 5 that allegedly enabled codebase analysis and software-flaw identification. Anthropic pushed back, calling the capability “widely available” from other frontier models, including OpenAI’s GPT-5.5. A U.S. official confirmed the letter to Bloomberg. The written directive cited national security without specifics.

The sequencing is the story. Fable 5, a guard-railed derivative of Mythos 5, shipped roughly three days before the directive arrived. Mythos 5 itself had been gated to about 50 vetted organisations under Project Glasswing. In March, the Pentagon had already designated Anthropic a “supply chain risk,” a label the company is contesting in federal court. Earlier in June, Anthropic confidentially filed for a public listing on the back of a funding round valuing it at $965 billion.

European capitals read the episode quickly. Speaking ahead of the G7 summit at Evian-les-Bains, Canadian Prime Minister Mark Carney warned of the cost of “over-reliance.” Former French minister Bruno Retailleau called Anthropic a “wake-up call” for Europe.

The structural read: the same safety-forward posture that earned Anthropic regulatory trust in Washington has now made it the test case for unilateral American export power over frontier models. The bill for that posture is being itemised in Brussels, Ottawa, and, eventually, an S-1.

Sources