Anthropic: The US government has requested the suspension of foreign national access to Fable5 and Mythos5 to ensure compliance and will disable the relevant models for all users

**Breaking: Anthropic Disables Fable 5, Mythos 5 Under U.S. Export Control Directive** June 13 — Anthropic confirmed today the U.S. government issued a national security-linked export control directive, ordering immediate suspension of all foreign national access to its Fable 5 and Mythos 5 models—whether located inside or outside the U.S., including its own foreign national employees. To comply, Anthropic must block all user access to the two models. Other Anthropic AI models remain unaffected. The directive arrived at 17:21 ET with no specific details on the national security concerns driving the order. The government’s apparent concern centers on a narrow, non-generalizable technique to spot minor, pre-known vulnerabilities in a demo environment—a capability already available from other public AI models, no jailbreak needed. Anthropic notes tools like OpenAI’s GPT-5.5, used daily by cybersecurity defenders to secure systems, can perform the same task. Pre-launch, Fable 5 underwent thousands of hours of red team testing with the U.S. government, UK AISI, private third parties, and internal teams. The tests found Fable 5’s security measures far stronger than any prior deployed AI, and no widespread jailbreak method was uncovered. The government only provided a verbal hint of this narrow potential issue, not concrete proof. Anthropic will comply with the directive to pull Fable 5 and Mythos 5 access, but disputes the decision to take a commercial model used by millions offline over a narrow, unproven jailbreak risk. The firm warns such a standard would block all leading AI providers from launching new models, calling for a transparent, fact-based government process for AI security. Anthropic labels the action a misunderstanding and says it’s working to restore access as soon as possible.