At 5:21 p.m. ET Friday, a letter from Commerce Secretary Howard Lutnick landed in Dario Amodei's inbox barring foreign-national access to Fable 5 and Mythos 5. By Friday night, Anthropic had disabled both models for every customer on earth, citing an inability to filter users by nationality in real time. Fable 5 had been live for three days.

The model Vals AI's benchmarks rated the most capable available to the public, deployed commercially to hundreds of millions of people and to the roughly 50 vetted organizations inside Project Glasswing, is now dark. So is Mythos 5. The rest of the Claude family remains online.

Anthropic's public statement is unusually blunt for a company that spends most of its press cycles cultivating the regulator class. "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 company wrote. "If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers." The triggering jailbreak, per Anthropic, was a researcher asking the model to audit a codebase, a use case that Amazon, Apple, Google, Microsoft, and CrowdStrike pay for daily.

The structural irony, raised first by TechCrunch, writes itself. Anthropic has built its commercial identity on dangerous-capability marketing: the thousands of hours of red-teaming, the UK AI Safety Institute partnership, the carefully telegraphed worry about what its own models can do. Sam Altman called the Mythos rollout "fear-based marketing" to Ashlee Vance in April. The Trump administration appears to have read the marketing material and acted on it.

It also fits a pattern. Earlier this year the Department of Defense designated Anthropic a supply-chain risk and barred defense contractors from using Claude in military work; that litigation is ongoing. The June executive order on pre-deployment testing gave Commerce the procedural hook Lutnick used Friday. And White House AI adviser David Sacks, who has publicly opposed a licensing regime, now presides over what's functionally a license revocation by letter, applied to one company, three days after launch, on the basis of a code-audit prompt.

Anthropic is calling the order a misunderstanding. Whether it's or isn't, the company spent years arguing that frontier models were uniquely dangerous and required government attention. It got the attention. The lesson the rest of the industry is absorbing, in real time, is that telling Washington your product is too powerful to release is a marketing posture with a settlement date.

Anthropic pulls Fable 5 and Mythos 5 globally after Commerce export-control order over alleged jailbreak

Sources