Anthropic is shutting down Fable and Mythos AI models following a US government export ban

The directive applies not only to people outside the US, but to all foreign nationals in the US, including Anthropic’s own non-US employees.

Given the scope of the directive, Anthropic argued that it had no choice but to disable the models for all users. He clarified that access to less powerful Claude models, including the latest Claude Opus 4.8, was not affected.

“We apologize to our customers for this disruption,” Anthropic wrote in a post on X. “We believe this is a misunderstanding and are working to restore access as soon as possible.”

Anthropny said in a blog post that he received the directive at 5:21 PM ET. He said the letter he received “does not contain specific details” of the government’s national security concerns.

But the company also said officials told Anthropic that the government made the decision after learning of a method to circumvent Fable 5’s security measures. Those security measures were designed to prevent users from accessing the powerful cybersecurity capabilities of Mythos, the core artificial intelligence model on which Fable 5 is built.

Anthropic said it believed the jailbreak cited by the government was a narrow one that would only expose Mythos’ cybersecurity capabilities in one specific case, rather than a universal one that would defeat all of Fable 5’s security measures. It also said it believes the same jailbreak could be used to gain similar capabilities from other public models, including GPT-5.5 OpenAI, which are not subject to similar national security export controls.

“We disagree that the discovery of a narrow potential jailbreak should be grounds for revoking a commercial model deployed to hundreds of millions of people,” Anthropic wrote in its blog. “If this standard were applied across the industry, we believe it would essentially stop the rollout of all new models for all frontier model suppliers.

“As we’ve said publicly, we believe the government should be able to block dangerous deployments through a legal process that is transparent, fair, clear, and based on technical facts. This action does not meet those principles,” Anthropic said.

Artificial intelligence industry insiders and policy experts reacted with disbelief to the unprecedented US directive.

Some saw the move as another attempt by the Trump administration to punish Anthropic. U.S. President Donald Trump in February ordered all federal agencies to stop using Anthropic’s models after the company refused to agree to preferential terms of the Pentagon’s AI vendor contract, which stipulated that any purchased AI models could be used “for any lawful purpose.” Anthropic sought an exemption from using its models for autonomous weapons systems or mass home surveillance.

In early March, the Pentagon declared Anthropic a “supply chain risk,” requiring the U.S. military to stop using their models and banning defense contractors from using them for government contracts. Anthropic is challenging that designation in federal court.

Anthropic confidentially submitted public list earlier this month. A recent funding round valued the company at $965 billion. The government’s decision to impose export controls could dampen investor enthusiasm for Anthropic’s IPO, raising doubts about whether it will be able to stay on the cutting edge of AI model development if the government continues to single out its models for various restrictions.

Several of Trump’s key technology policy advisers, most notably former AI and cryptography czar David Sachs and the Pentagon’s undersecretary of defense for research and technology Emil Michael, have publicly attacked Anthropic and its executives. Sachs accused Anthropic of being “woke” and “leftist” and engaging in a “sophisticated regulatory strategy based on fear mongering.”

Dean Ball, an AI policy expert who spent a short time in the Trump administration but has been highly critical of Anthropic’s recent decisions, said on X that “I can’t tell if this is an anti-Anthropic law specifically or an extreme national security activity. Regardless, it’s just a cartoon.”

He added: “An administration that thinks we *should* export advanced AI chips to China, that also wants to ban… Britain (and every other non-American on Earth)… from using our best models? I’m speechless.”

But others said Anthropic is simply reaping what it has sown. When Anthropic first debuted its Mythos model, the company claimed the model was too dangerous for mass release. When it released its Fable 5 model, which is based on Mythos, the company highlighted the security measures it put in place to prevent users from accessing all of Mythos’ features.

“If you describe your product as ammunition in every press release, eventually the government will take your word for it,” Peter Girnus, a cybersecurity researcher who describes himself in his X handle as a “cyber populist.” said at X. “They wrote the legal predicate themselves and called it a brand.”

Girnus also noted that previous government attempts to restrict software exports, such as an attempt to impose restrictions on powerful encryption techniques in the 1990s, have largely failed.

Gary Marcus, a frequent critic of the AI ​​industry, said in a social media post, he believes the government’s actions make no sense, especially given his oft-stated position that the US should outpace China in developing powerful artificial intelligence systems. The national security directive is likely to persuade many Chinese-born AI researchers currently working at labs such as Anthropic and OpenAI to return to China, he said. He added that this would make investors question whether US AI companies are a safe bet, given the apparently capricious nature of the Trump administration’s AI policy.

Interesting, Bal pondered that some people concerned that artificial intelligence poses a threat to human existence, including perhaps Anthropic’s own AI security staff, may welcome the government’s decision. That’s because it could slow down the development of AI, which is what these AI safety advocates have been pushing for.