Anthropic is disabling its Fable 5 and Mythos 5 models for all customers to comply with a U.S. government export-control directive that, citing national security, bars any foreign national from accessing them. Anthropic said the government did not detail its concern but that it appears to involve a narrow jailbreak, essentially asking the model to find and fix software flaws in a codebase, a capability Anthropic says other models, including OpenAI's GPT-5.5, can already perform. The company said it is complying but disagrees that a narrow jailbreak should justify recalling a model used by hundreds of millions, calling the move a misunderstanding and saying it is working to restore access, while its other models stay available.
