The U.S. government has mandated Anthropic to immediately disable access to its two most powerful AI models, Claude Fable 5 and Claude Mythos 5, due to national security concerns. Anthropic confirmed it has complied with the directive, though it expressed disagreement with the decision. The order, received on Friday at 5:21 pm ET, requires the company to shut down both models for all users worldwide, not just foreign nationals. Access to other models remains unaffected.
Mythos 5, Anthropic’s most capable AI model, was previewed in early April and has been tightly restricted due to its ability to identify security vulnerabilities in major operating systems and web browsers. The company launched Project Glasswing, a controlled program sharing Mythos with about 50 vetted organizations for defensive cybersecurity work. Fable 5, released three days ago, is a version of Mythos with guardrails to block responses in high-risk areas, making it suitable for general release. It was immediately the most capable AI model available to the public, according to Vals AI benchmark tests.
Anthropic argues that the government’s concern about a potential jailbreak of Fable 5 is based on verbal evidence of a narrow, non-universal jailbreak. The company claims this level of capability is already available in other publicly accessible models, including OpenAI’s GPT-5.5, and is routinely used by cybersecurity professionals for defensive purposes. Anthropic emphasized that its strongest safeguards operate through independent classifier systems, which function separately from the model itself. Despite these arguments, the government proceeded with the shutdown, and Anthropic is not hiding its frustration. Source: techcrunch
