Anthropic disables Fable and Mythos AI models after U.S. Commerce Department order

Anthropic announced on Friday night that it has disabled its Fable 5 and Mythos 5 large-language models for all customers after receiving a directive from the U.S. Commerce Department that subjects the models to export-control restrictions. The company said the action was necessary to ensure immediate compliance, while access to its other models remains unchanged.

The Commerce Department’s concern stems from reports of a “jailbreak” that could bypass classifier-based safeguards and allow the models to generate content related to cybersecurity, chemistry and biology. An administration official, cited by Axios, said the pause would give the national-security apparatus time to harden defenses, a process that could be completed within weeks.

Anthropic said the government provided only “verbal evidence of a potential narrow, non-universal jailbreak” involving the review of a specific codebase for software flaws. The company noted that the jailbreak appears to target “minor” and “relatively simple” software vulnerabilities and that comparable capabilities exist in other publicly available models.

"We are complying with the government’s legal directive and are removing access to Fable 5 and Mythos 5 for all users," the company wrote. "However, 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. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers."

The move follows an executive order signed earlier this month by President Trump urging AI developers to submit to voluntary government security testing. Anthropic apologized to customers for the disruption, describing it as the result of a “misunderstanding,” and said further details would be provided within 24 hours.