Anthropic suspends Fabel and Mythos after US export controls
Anthropic suspended access to Fabel 5 and Mythos 5 for all users after the U.S. Commerce Department imposed export controls citing a potential jailbreak risk.
Anthropic suspended access to its Fabel 5 and Mythos 5 models for all users on Friday after the U.S. Department of Commerce imposed export controls that bar foreign nationals from using those systems.
The Commerce Department announced the restrictions following a report from Amazon researchers who identified a possible method to bypass safeguards in Fabel. The department described the finding as a potential jailbreak risk.
Anthropic complied with the legal directive and removed access to the named models for all accounts, saying regulators presented only verbal evidence of a “potential narrow, non-universal jailbreak” and disputing the scope of the restriction.
In a statement, Anthropic wrote, “We disagree that the finding of a narrow potential jailbreak should be a cause for recalling a commercial model deployed to hundreds of millions of people.” The company added that applying the same standard industry-wide “would essentially halt all new model deployments for all frontier model providers.”
A jailbreak is a technique that tricks an AI into bypassing built-in safety filters and guardrails. Anthropic noted that similar vulnerabilities can appear in other advanced models and that security teams regularly test and defend against such exploits.
The export-control directive specifically prevents foreign nationals from accessing the named models. Anthropic said it removed access for all users rather than offer a split service to comply with the legal order.
People familiar with the matter indicated Amazon first shared its findings with Anthropic, and that Amazon CEO Andy Jassy raised questions about frontier-model capabilities with U.S. officials.
The suspension follows the Pentagon naming Anthropic as a supply-chain risk. Senior government officials and Anthropic leadership have also publicly clashed over AI regulation, the use of technology in domestic surveillance and the development of autonomous weapons.
Anthropic did not provide a timeline for restoring access. The company said it will continue to cooperate with government directives while contesting the factual basis for the restriction.








