Anthropic was forced to pull its two most powerful AI models offline late Friday after the US Commerce Department issued a national security directive barring the company from distributing them to any foreign national — a sweeping order that left the company with no practical choice but to disable the models entirely for all users worldwide.
The directive targeted Fable 5 and Mythos 5, Anthropic’s newest and most capable models. It covered not just users outside the United States but any foreign national anywhere — including Anthropic’s own non-citizen employees working inside the country. Access to the company’s less powerful Claude models, including its latest Claude Opus 4.8, was not affected.
Anthropic received the directive at 5:21pm Eastern Time. The letter provided no specific details of the government’s national security concern. Company officials were subsequently told verbally that the decision followed the discovery of a technique to bypass Fable 5’s safeguards — protections designed to prevent users from accessing the advanced cybersecurity capabilities of Mythos, the underlying model on which Fable 5 is built.
“We apologise for this disruption to our customers,” Anthropic wrote on social media. “We believe this is a misunderstanding and are working to restore access as soon as possible.”
What Anthropic Says — and Why It Is Pushing Back
Anthropic’s response was pointed and detailed. The company said it believes the jailbreak cited by the government is a narrow one — capable of unlocking Mythos’s cybersecurity capabilities in only one specific scenario rather than defeating all of Fable 5’s safety systems broadly. It also argued that the same technique can be used to elicit similar capabilities from other publicly available frontier models, including OpenAI’s GPT-5.5, which are not subject to equivalent export controls.
“We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people,” the company wrote in a blog post. “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
Anthropic was clear about what it believes proper process should look like. “We believe the government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not adhere to those principles.”
ALSO READ: SpaceX Is Now Public at $1.77 Trillion — Here’s What Investors Need to Know About Buying SPCX
A Pattern of Government Pressure on Anthropic
The directive does not exist in isolation. In February, President Trump ordered all federal agencies to cease using Anthropic’s models after the company refused to agree to Pentagon contract terms that would have allowed its AI to be used “for any lawful purpose” — Anthropic had sought exemptions from deployment in autonomous weapons systems and mass domestic surveillance programmes. The Pentagon subsequently designated Anthropic a “supply chain risk” in March, requiring the US military and defence contractors to stop using its models entirely. Anthropic is challenging that designation in federal court.
Several senior Trump technology advisers have been publicly hostile toward the company. Former AI and crypto policy czar David Sacks has called Anthropic “woke” and “leftist” and accused it of a “sophisticated regulatory capture strategy based on fear-mongering.”
The timing creates significant complications for the company’s financial future. Anthropic filed confidentially for a public listing earlier this month after a recent funding round valued it at $965 billion. A government that has repeatedly singled out its models for restrictions is not an easy backdrop for an IPO, and investors are likely to scrutinise the political risk that has accumulated around the company.
How the AI Industry and Policy World Reacted
Reactions from technology policy experts ranged from alarm to something approaching dark humour. Dean Ball, an AI policy researcher who briefly served in the Trump administration and has been critical of its recent approach to Anthropic, described the directive on social media as “simply cartoonish.” He pointed out the contradiction of an administration willing to export advanced AI chips to China while simultaneously banning British and other allied-nation users from accessing a commercial AI model. “I have no words,” he wrote.
Gary Marcus, a prominent critic of the AI industry, said the action made little strategic sense given the US government’s stated position that America must stay ahead of China in frontier AI development. He argued the directive would likely push Chinese-born AI researchers currently working at American labs to return to China, and would cause investors to question whether US AI companies represent a safe long-term bet.
A cybersecurity researcher offered a more unsympathetic reading of Anthropic’s situation, arguing the company had partly written its own predicament by describing Mythos in terms that invited exactly this kind of regulatory response. “If you describe your product as a munition in every press release, eventually a government takes you at your word. They wrote the legal predicate themselves and called it a brand.”
Ball also noted an irony in the broader AI safety community’s likely reaction — that some within Anthropic itself, and others who believe AI poses existential risks, might quietly welcome the decision if it has the effect of slowing the pace of AI development. It is, he suggested, a strange kind of victory if it comes to that.
Stay informed. Subscribe to the JournalTodays Newsletter for the latest AI news, technology policy, and business coverage delivered straight to your inbox.
