Anthropic to Shut Down Fable 5 and Mythos 5 Models Following U.S. Export Control Order

Anthropic announced it will disable access to its latest AI models, Fable 5 and Mythos 5, after receiving a U.S. government export-control letter that cited national-security risks. The company said the notice, received around 5:21 p.m. ET on Friday, applies to all foreign individuals and entities, including foreign-national employees, and requires the models to be turned off for everyone to ensure compliance.

The letter did not detail the specific security concerns, but Anthropic believes the government is worried about a potential "jailbreak" of Fable 5. The firm argues the vulnerability is limited, not universal, and can be reproduced with publicly available models.

The decision follows a broader dispute between Anthropic and the Trump administration over AI safety and government oversight. In February, the Pentagon labeled the startup a supply-chain risk after Anthropic sought usage limits for defense applications, leading to lawsuits that remain pending.

A White House spokesperson did not immediately respond to requests for comment. The Pentagon’s chief information officer posted on X that certain priorities outweigh revenue concerns.

Anthropic said it will comply with the order, though it disagrees with the government’s assessment, and did not specify the exact timing for disabling the models. Access to the company’s other AI models will continue unchanged.