Washington Orders Anthropic to Cut Off Two Models From All Foreign Nationals

On June 12, the United States government issued Anthropic an export-control directive to suspend all access to its Fable 5 and Mythos 5 models by any foreign national, anywhere in the world, including Anthropic's own foreign-national employees. Unable to apply that restriction selectively, the company disabled both models for every customer, three days after they launched on June 9.

The government's stated rationale, as Anthropic describes it, rests on national-security authorities but came with little detail. Anthropic's understanding is that officials believe they found a method of bypassing Fable 5's safeguards, a narrow jailbreak that consists essentially of asking the model to read a codebase and fix its software flaws. The company says it received only verbal evidence and disputes that a narrow, non-universal jailbreak justifies recalling a model deployed to hundreds of millions of people.

The episode marks an escalation in how governments treat frontier models. Export controls have until now targeted hardware, the graphics processing units (GPUs) and lithography tools that underpin AI systems. Extending them to application programming interface (API) access to a specific model, and to who may use it rather than what may be sold, treats the model itself as a controlled munition. In practice, it places the model under the kind of export restrictions normally applied to weapons. For any engineer building on a US frontier API, it introduces a new category of platform risk. A model can be disabled without warning by government order, with no contractual recourse.