US Government Forces Anthropic to Suspend Claude Fable 5 and Mythos 5

Anthropic has abruptly disabled access to its newly launched Claude Fable 5 and Mythos 5 models following a U.S. government export control directive. The order restricts access for all foreign nationals, forcing the AI company to suspend the models for all customers globally to ensure full compliance.



The company said it received the directive at 5:21 p.m. ET on Friday. Citing national security authorities, the government ordered the suspension but did not provide specific details about its concerns. Anthropic said it believes the issue centers on a method for bypassing, or "jailbreaking," the models. According to the company, the concern stems from a narrow technique where the AI is asked to read a specific codebase and fix software flaws.


Anthropic strongly disagrees with the suspension. After reviewing a report it believes served as the basis for the directive, the company stated that the capabilities in question are already widely available from other frontier models, including OpenAI's GPT-5.5. This regulatory crackdown arrives shortly after security researchers demonstrated how Anthropic's AI could be used to bypass advanced macOS protections, an event that intensified scrutiny of advanced AI systems in Washington.

The sudden shutdown comes just days after Anthropic launched Fable 5 and Mythos 5. Introduced with a focus on software engineering and coding tasks, the models quickly attracted attention for their advanced coding and security-analysis capabilities. Before release, Anthropic spent thousands of hours red-teaming the software with U.S. and UK government agencies. The company maintains that a universal jailbreak has not been discovered and that its security monitoring strategy keeps the models' risks comparable to existing industry alternatives.

For now, Anthropic is complying with the legal order while working to restore access. The company warned that pulling a commercial model over a narrow jailbreak sets a precedent that could effectively halt all future frontier AI deployments across the entire tech industry. Access to older models like Claude Opus remains unaffected.

Read More