BLOG
Anthropic Suspends Access to Fable 5 and Mythos 5 Following US Government Directive
Anthropic has suspended access to its newly launched Claude Fable 5 and Claude Mythos 5 models after receiving a US government export control directive on June 12, 2026.
According to Anthropic, the directive requires the company to suspend access to both models for any foreign national, whether located inside or outside the United States, including foreign national employees of Anthropic. The company stated that the net effect of the order is that access to Fable 5 and Mythos 5 must be disabled for all customers to ensure compliance with the directive. Access to Anthropic's other AI models remains unaffected.
What Are Fable 5 and Mythos 5?
Claude Fable 5 and Claude Mythos 5 are Anthropic's most advanced AI models to date. Both share the same underlying model and demonstrate state-of-the-art performance across software engineering, reasoning, scientific research, vision, and long-context tasks.
Fable 5 is the public version designed for general use. Because of its advanced capabilities, Anthropic deployed it with strong safety safeguards, particularly around areas such as cybersecurity. In some cases, requests on sensitive topics are automatically handled by a less capable model to reduce potential misuse.
Mythos 5 is a restricted-access version of the same model available through Project Glasswing to a limited group of trusted cyber defenders and infrastructure providers. It operates with some of Fable 5's safeguards removed and is intended for advanced cybersecurity and research applications.
Why Did the Government Issue the Directive?
Anthropic stated that the government did not provide detailed information about its national security concerns in the directive. However, the company's understanding is that the government became aware of a technique that could potentially bypass some of Fable 5's safeguards.
After reviewing demonstrations of the technique, Anthropic said it appeared to identify a small number of previously known and relatively minor software vulnerabilities. The company also stated that similar findings can be produced by other publicly available AI models without requiring any safeguard bypass.
Anthropic further noted that the reported technique is not considered a universal jailbreak. According to the company, no testers have yet discovered a method capable of broadly bypassing Fable 5's safeguards across a wide range of capabilities.
Anthropic's Response
Anthropic emphasized that Fable 5 underwent extensive safety testing before release. The company worked with US government teams, the UK's AI Safety Institute, private organizations, and internal red teams for thousands of hours to evaluate the model's safeguards.
The company has long maintained that perfect jailbreak resistance is not currently achievable for any frontier AI model. Instead, Anthropic says it adopted a defense-in-depth approach designed to make bypass attempts narrow, difficult, and detectable through monitoring systems.
Anthropic stated that it has not received evidence showing that the reported technique resulted in harmful outcomes. It also said the examples reviewed so far do not demonstrate capabilities unique to Mythos 5 and are comparable to tasks already performed by other leading AI systems.
Looking Ahead
While Anthropic is complying with the government's directive, the company disagrees with the decision to suspend access based on what it describes as a narrow potential jailbreak. Anthropic argues that governments should have the authority to block unsafe AI deployments, but that such actions should occur through a transparent, fair, and technically grounded process.
The company says it believes the current situation is a misunderstanding and is working to restore access to Fable 5 and Mythos 5 as soon as possible.
Source: Anthropic News 1, Anthropic News 2