Anthropic Disables Claude Fable 5 and Mythos 5 Following US Government Export Control Directive
Anthropic has disabled its Claude Fable 5 and Mythos 5 AI models for all customers following a US government export control directive issued on June 12, 2026. The directive cited national security authorities, suspending access by any foreign national. Anthropic complied by shutting down the models globally, as it could not filter users in real-time. The company, however, disputes the rationale, suggesting the government's concern stems from a 'narrow jailbreak' and calls it a misunderstanding. All other Anthropic models, including Claude Opus 4.8, remain available.
Anthropic has disabled its two most capable artificial intelligence models, Claude Fable 5 and Claude Mythos 5, for all customers. This action followed a US government export control directive issued on June 12, 2026, which cited national security authorities.
The order specifically named Claude Fable 5 and Claude Mythos 5, both of which had launched just three days prior on June 9. The directive suspended access for any foreign national, both inside and outside the United States. This scope also included Anthropic's own foreign national employees. Due to an inability to filter foreign nationals from US users in real time, Anthropic disabled both models for everyone to ensure compliance. Access to all other Anthropic models, such as Claude Opus 4.8, remains unaffected.
Commerce Secretary Howard Lutnick sent the order to Anthropic CEO Dario Amodei. While the letter did not specify the exact national security concern, Anthropic's interpretation suggests the government may have learned of a method to "jailbreak" Fable 5. A jailbreak bypasses a model's safeguards to elicit otherwise blocked behaviors.
Anthropic reviewed a demonstration tied to what it described as a few minor vulnerabilities. The company stated that similar flaws are found in other public models without a bypass and that the demonstrated capability is widely available elsewhere, citing OpenAI's GPT-5.5 as a comparable example. The Commerce Department reportedly acted after another company claimed to have jailbroken Mythos, raising concerns among officials about potential national security risks. The administration had previously sought to delay the models' launch, which Anthropic declined, leading to the export control letter.
While complying with the directive, Anthropic is openly disputing the rationale, calling the situation a likely misunderstanding. The company argues that the cited jailbreak is narrow and non-universal, questioning whether such a technique should justify a full recall. Anthropic warned that such a standard could halt new deployments across the industry. Claude Fable 5, a Mythos-class model, employs separate AI systems called classifiers to detect potential misuse, with flagged queries falling back to Claude Opus 4.8.
According to Marktechpost, Anthropic implemented a "defense in depth" strategy and red-teamed its safeguards extensively before launch, involving partners like the US government and the UK AISI. An external bug bounty program found no universal jailbreak in over 1,000 hours of testing, and Anthropic stated no tester has found one on production tasks yet.
