Anthropic Pulls Fable 5, Mythos 5 After US Directive Cites "Potential Narrow, Non-Universal Jailbreak" 🔓

Anthropic has suspended access to its Fable 5 and Mythos 5 AI models following a US government export control directive received at 5:21 pm ET on Friday. The directive ordered the company to halt all access to the two models by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. Anthropic said it abruptly disabled the models for all users in order to ensure compliance. Other Anthropic models, including Opus 4.8, are not affected.

The suspension comes just days after Anthropic released Fable 5 and Mythos 5, two AI models built on top of Mythos Preview, a general-purpose language model the company had previously said identified thousands of vulnerabilities in critical software. Fable 5 was released on Tuesday to the general public with built-in safeguards designed to prevent misuse, while Mythos 5 was made available only to approved cybersecurity partners. Anthropic said the government did not provide specific details about the alleged threat but indicated authorities were concerned about a possible "jailbreak" method capable of bypassing the models' safeguards.

Anthropic pushed back on the characterization, stating that across thousands of hours of testing no testers had found a "universal jailbreak" capable of broadly bypassing the model safeguards. "To date, the government has only given us verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws," the company said. The company added that it believes the models remain in compliance with the spirit of the government's stated concerns.