Anthropic removes Mythos and Fable after U.S. order
Anthropic AI models go offline after U.S. access limits and jailbreak concerns See why the move raises AI safety, export-control, and global access questions
Anthropic has taken its Mythos and Fable 5 AI models offline worldwide after a U.S. directive required the company to block access for nonU.S. citizens. The company said the restriction followed concerns about a reported jailbreak tied to the models.
According to reporting cited in the article, the move may have been influenced in part by worries that foreign actors could access the models, including possible Chinalinked interest. Anthropic said it was given only verbal evidence about the issue and argued that similar concerns can also be found in other leading AI systems.
The decision highlights the growing tension between AI safety, export controls, and access to frontier models. It also adds to Anthropic’s ongoing relationship with U.S. regulators as the company continues to advocate for stricter AI oversight.