Anthropic said it has suspended access to its Fable 5 and Mythos 5 AI models after receiving a US government export control directive citing national security authorities.
The company said it received the order at 5:21 p.m. ET on June 12 and was required to disable access to the two models for all customers, including foreign nationals inside and outside the United States and foreign-national Anthropic employees. Access to the company’s other AI models remains unaffected.
“The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national,” Anthropic said in a statement.
Anthropic said the government did not provide specific details about its national security concerns. According to the company, its understanding is that officials had identified a method of bypassing, or jailbreaking, Fable 5.
The company said it reviewed a demonstration of the technique and found that it exposed a small number of previously known, minor vulnerabilities. Anthropic added that other publicly available AI models could identify the same vulnerabilities without requiring a jailbreak.
“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,” the company added.
Anthropic said extensive testing of Fable 5 had been conducted before launch, involving the US government, the UK’s AI Safety Institute, third-party organisations, and internal teams. The company said the testing found Fable’s safeguards to be more effective than those of earlier deployed models and that no tester had identified a “universal jailbreak” capable of broadly bypassing the model’s protections.
The company acknowledged that no AI provider currently appears capable of achieving complete jailbreak resistance and said all industry safeguards remain vulnerable to narrower jailbreak techniques under certain circumstances.
Anthropic said its approach relied on a “defence in depth strategy” that combines safeguards, monitoring and customer data retention policies to detect and mitigate jailbreak attempts.
“We stand by this defence-in-depth strategy. It reduces the risks posed by Fable, making them comparable to the risks of existing models already deployed across the industry,” the company said.
The company said it had not received evidence of a jailbreak that resulted in harm and that the disclosed techniques either generated benign responses or identified minor software issues. Anthropic added that the government had provided only verbal evidence of a potential non-universal jailbreak involving requests to review code and identify software flaws.
Anthropic further said it had reviewed a report it believes formed the basis of the directive and found that the capability described was “widely available from other models (including OpenAI’s GPT-5.5), and is used every day by the defenders who keep systems safe.”
While complying with the order, Anthropic criticised the decision to recall a commercial AI model over what it described as a narrow potential 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.
Anthropic said governments should have the authority to block unsafe AI deployments, but argued that such actions should be carried out through “a statutory process that is transparent, fair, clear, and grounded in technical facts.”
The order comes days after Anthropic proposed a framework for government oversight of frontier AI models. In a recent blog post, CEO Dario Amodei called for mandatory third-party testing of advanced systems for risks related to cybersecurity, biological weapons, loss of control, and automated research capabilities. “The government should have the power to block or deter deployment of the model if it is determined, in light of third-party assessment, to present unacceptable risks,” Amodei wrote.
Meanwhile, the company said it is working to restore access to Fable 5 and Mythos 5 and believes the government’s action is based on a misunderstanding.
ALSO READ: Claude Mythos Has Changed the Vulnerability Curve. Can Defenders Keep Up?
