New Delhi: US-based AI company Anthropic disabled access to its two most advanced models—Fable 5 and Mythos 5—for all users outside the US and foreign nationals even in the US Friday evening. This came after the US government issued an export control directive barring foreign nationals from the tools over national security concerns—the first time a leading AI company has taken a publicly deployed model offline due to federal government intervention.
Commerce Secretary Howard Lutnick sent a letter to Anthropic CEO Dario Amodei Friday, stating that Mythos 5 and Fable 5 would be subject to export controls in any location outside the US and to all foreign persons within the country. The restrictions apply even to Anthropic’s own foreign employees. Any export, re-export, or transfer of the models within the US now requires an individual licence.
Anthropic said in a statement that it received the directive at 5.21 pm ET and that the letter offered no specifics. “The letter did not provide specific details of its national security concern,” the company wrote. Because it could not selectively block foreign nationals in real time across a shared cloud service, it disabled both models for every customer globally. All other Anthropic models remain unaffected.
The timing is notable. The shutdown comes just days after Anthropic announced Fable 5 and Mythos 5, two models the company touted as state-of-the-art across industry benchmarks. Fable 5 was the first Mythos-class model made available to the general public, built with safeguards designed to block responses in high-risk areas such as cybersecurity. Mythos 5, in contrast, is the same underlying model with those safeguards removed, reserved exclusively for vetted cybersecurity defenders and critical infrastructure operators.
Both models grew out of Claude Mythos Preview, a system initially limited to a small group of companies and research partners through Project Glasswing, a collaboration between Anthropic and the US government. Participants used the model to find and repair software vulnerabilities.
Mozilla alone said it resolved hundreds of issues as a direct result. Project Glasswing brought together firms including AWS, Apple, Cisco, Google, Microsoft, and Nvidia for that purpose. When Fable 5 launched this week, Mythos 5 remained restricted to Project Glasswing participants while Fable was opened to the public.
An administration official told Axios, the Commerce Department acted after another company claimed it was able to jailbreak Mythos, alarming the administration about possible national security risks. The administration had tried to get Anthropic to pause the release of the models but was unsuccessful, prompting the export control letter.
Anthropic pushed back on the government’s reasoning. The company said it had reviewed a demonstration of the technique and found it surfaced only minor, previously known vulnerabilities.
“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,” Anthropic stated.
The company said it believed the same jailbreak could be used to elicit similar capabilities from other publicly available models, including OpenAI’s GPT-5.5, which is not subject to similar export controls.
Anthropic said it had worked with the US government, the UK AI Security Institute, and multiple private organisations to stress-test Fable’s safeguards for thousands of hours before launch, and that no testers had found a universal jailbreak.
“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 said, adding that applying such a standard industry-wide “would essentially halt all new model deployments for all frontier model providers”.
The move fits a pattern of US efforts to manage perceived risks from advanced AI, even as critics argue such restrictions may slow domestic innovation without meaningfully addressing global proliferation. Jailbreaking remains a common dynamic across the industry, suggesting no single model holds a monopoly on concerning capabilities.
Anthropic said it is complying with the directive while calling the situation “a misunderstanding” and said it is working to restore access. No timeline has been given. “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,” the company stated. “This action does not adhere to those principles.”
(Edited by Viny Mishra)
Also read: AI is now building itself, getting out of Human control. Anthropic wants everyone to pause