Anthropic has confirmed it has taken down two AI models it launched earlier this week, Fable 5 and Mythos 5, to comply with an export control directive citing national security concerns that the US government released late Friday evening on June 13th.

This is just the latest blow in the ongoing conflict between Anthropic and US President Trump's administration. The order requested that Anthropic suspend access to "any foreign national, whether inside or outside the United States, including foreign national Anthropic employees." However, Anthropic has pulled access to all customers to ensure compliance with the directive. In its blog post about the move, Anthropic says that it received a letter from the US government at 5:21pm ET regarding the government directive, and that it did "...not provide specific details of its national security concern."

"Our understanding is that the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5," the company added in the blog post. "We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass."

Anthropic says that it has safeguards in place to prevent bad actors

Earlier this year, Trump designated Anthropic a "supply chain risk"

Anthropic argues that it has strong safeguards in place that notably reduce the possibility of Claude Fable 5 being used by bad actors, emphasizing that the "jailbreak" the US government uncovered is very limited.

"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," writes Anthropic in the blog post.

Earlier this year, Trump's Department of Defense labeled Anthropic a "supply chain risk" after the AI company limited the US military's use of its technology, citing ethical concerns. This designation barred government agencies from using Anthropic's technology. The company then filed lawsuits against the US government.

On Tuesday, June 9th, Anthropic released Claude Fable 5, an updated version of its Mythos AI model with new safeguards related to biology, chemistry, and cybersecurity. This model, created through a collaboration with the US government, was initially available in a limited preview in April. The overarching goal was to curb concerns surrounding Anthropic's AI technology being used by bad actors.

It's unclear when or if Claude Fable 5 and Mythos 5 will go back online. Anthropic is likely working to patch the vulnerability that the US government claims to have uncovered, potentially leading to the return of the AI models.

XDA has reached out to Anthropic and will update this story with more information as it becomes available.