US orders shutdown of Anthropic’s Fable 5, Mythos 5 over security concerns

The US government has ordered the immediate suspension of access to Fable 5 and Mythos 5, advanced artificial intelligence models developed by Anthropic, citing national security concerns.

The directive, issued on June 12 at 5:21 pm Eastern Time, applies to all foreign nationals globally, including Anthropic’s own non-US employees, effectively forcing the company to disable the models for all users to ensure compliance.

“The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance,” Anthropic said, adding that “access to all other Anthropic models will not be affected.”


The company stated that the government did not provide detailed evidence outlining the specific national security concerns behind the decision. However, officials indicated that the directive is linked to concerns over a potential method to bypass, or “jailbreak,” the safeguards built into Fable 5.

The company further said, “This action does not adhere to those principles,” referring to the need for a process that is “transparent, fair, clear, and grounded in technical facts.” It added, “We believe this is a misunderstanding and are working to restore access as soon as possible.”

According to Anthropic, the identified technique involves prompting the model to analyse codebases and detect software vulnerabilities. The company said it reviewed a demonstration of this method and found that it exposed only a limited number of previously known and minor issues. It added that similar capabilities are widely available in other publicly accessible AI systems and are routinely used by cybersecurity professionals.

Anthropic emphasised that extensive pre-release testing of Fable 5 was conducted in collaboration with US and UK government bodies, independent organisations, and internal teams. These efforts included thousands of hours of red-teaming to evaluate safeguards.

The company reported that no universal jailbreak method defined as one capable of broadly bypassing protections has been identified.

The Dario Amodei-led firm reiterated its “defense in depth” approach, which combines layered safeguards with monitoring mechanisms, including a 30-day data retention policy aimed at detecting and mitigating misuse.