Anthropic forced to disable Fable 5 after US directive

Anthropic has disabled access to Claude Fable 5 and Claude Mythos 5 after receiving a US government export control directive citing national security authorities.

The company said the directive requires it to suspend access to the models by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. Anthropic said the practical effect is that it must remove access to Fable 5 and Mythos 5 for all customers to ensure compliance. Access to its other models is not affected.

According to Anthropic, it received the directive on 12 June at 5:21 p.m. ET. The company said the order did not provide specific details of the national security concern, but that it understands the government believes it has become aware of a method for bypassing, or jailbreaking, Fable 5.

Anthropic said it reviewed a demonstration of the technique being used to identify a small number of previously known minor vulnerabilities. The company argued that those vulnerabilities appeared relatively simple and could also be identified by other publicly available models without requiring a bypass.

Anthropic said Fable 5 had been red-teamed before launch by its internal teams, the US government, the UK AI Safety Institute and third-party organisations. The company said no tester had found a universal jailbreak capable of broadly bypassing the model’s safeguards.

The company said it is complying with the directive but disagrees that a narrow potential jailbreak should justify recalling a commercial model. It also argued that applying such a standard across the industry could effectively halt new frontier model deployments.

Anthropic said governments should be able to block unsafe AI deployments through a transparent and technically grounded statutory process, but said the current action does not meet those principles. The company said it is working to restore access as soon as possible.

Why does it matter?

The case shows how national security and export-control powers can directly affect access to frontier AI systems after deployment. It raises a major governance question: when should governments be able to suspend access to advanced models, and what evidence, transparency and due-process safeguards should apply? The dispute also highlights the growing tension between frontier AI safety, commercial deployment, cross-border access and government intervention in dual-use technologies.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!