US export directive forces global shutdown of Claude Fable 5 and Mythos 5 — a narrow jailbreak claim takes down Anthropic's flagship models

Late Friday, Anthropic received a directive from the US government ordering it to suspend all access to Claude Fable 5 and Claude Mythos 5 for any foreign national — including foreign nationals employed by Anthropic inside the US. Because compliance would require identity-checking every user at the model tier, the net effect is a full global shutdown of both models.

The stated reason is a jailbreak claim. According to Anthropic's public statement, the government believes someone discovered a method to bypass Fable 5's safety guardrails by asking the model to read a specific codebase and identify software flaws. Anthropic says it was given "only verbal evidence of a potential narrow, non-universal jailbreak" — one the company does not believe warrants pulling a commercial model deployed to hundreds of millions of users.

Anthropic explicitly disputes the calibration: applying this standard uniformly would, in the company's reading, mean any narrow proof-of-concept jailbreak could trigger a full recall for any frontier model, "effectively halting frontier AI deployment across the industry."

The legal mechanism. The directive reportedly came from the US Commerce Department, citing export control authority under national security grounds — the same framework used to restrict semiconductor exports. According to Axios, the letter was signed by Commerce Secretary Howard Lutnick. This is the first confirmed application of export control authority to a commercial LLM in active deployment.

Impact for developers. If you were calling claude-fable-5 or claude-mythos-5 via the API as of June 12, those calls are now returning errors. All other Anthropic models — including Claude Opus 4.8, Sonnet 4.6, and the Haiku family — are unaffected. Anthropic says it is "working to resolve the misunderstanding and restore access." Monitor the Anthropic status page and statement page for updates.

Sources: Anthropic: Statement on the US government directive, TechCrunch: Anthropic's safety warnings may have backfired, Axios: Trump admin blocks foreign access, Bloomberg, The New Stack, NBC News