Anthropic Takes Two AI Models Offline to Comply With US Government Order

Anthropic announced it is taking down two AI models launched earlier this week—Claude Fable 5 and Mythos 5—to comply with an export control order it received from the US government on Friday afternoon, citing national security concerns. The unprecedented incident marks the latest flashpoint between Anthropic and the Trump administration.

According to the company, the order required suspending access for “any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.” However, to ensure full compliance, Anthropic removed access for all customers.

Earlier this year, the Trump administration’s Department of Defense designated Anthropic a “supply chain risk” after the company sought to set limits on how the US military could use its technology. That designation effectively barred government agencies and contractors from using Anthropic’s technology, prompting the company to sue the Trump administration.

On Tuesday, Anthropic publicly released Claude Fable 5, a version of its Mythos AI model with safeguards preventing it from answering questions about cybersecurity, biology, and chemistry. Prior to that public release—which Anthropic said was done in collaboration with the US government—the Mythos Preview AI model had a limited rollout in April. The goal was to allow companies and organizations to use its powerful cybersecurity capabilities to improve defenses and address concerns that the technology could be exploited by malicious actors to develop powerful hacking tools.

In a blog post on Friday, Anthropic said it received a letter from the US government at 5:21 p.m. ET. “The letter did not provide specific details of its national security concern,” Anthropic wrote. “Our understanding is that the government believes it has become aware of a method of bypassing, or ‘jailbreaking’ Fable 5.” The company added: “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 argued in its blog post that it had implemented strong safeguards to reduce the likelihood of misuse of Claude Fable 5. The company also claimed that the jailbreak found by the US government was narrow and would not make an attacker meaningfully more dangerous than they would be with another AI model. “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. “Our understanding is that one potential jailbreak was shared with the government.”

Spokespeople for the White House and the US Commerce Department did not immediately respond to WIRED’s request for comment.

Anthropic CEO Dario Amodei stated in a policy essay earlier this week that he and the company support a fair, structured, and transparent government process that would block the release of unsafe AI models. In Friday’s blog post, Anthropic argued that “this action does not adhere to those principles.”