Anthropic announced it was disabling access to the Fable 5 and Mythos 5 AI models for all users after receiving an export control directive from the US government, issued under national security authority without providing detailed explanation. The directive prohibits access to Fable 5 and Mythos 5 by any foreign nationals, both within and outside the US, including Anthropic's international employees. As a result, the company was forced to completely disable the models to comply. Access to other Anthropic models was not affected by this directive.
Anthropic believes the warrant was issued following the discovery of a jailbreaking technique for the Fable 5 model. Anthropic stated that it had examined the demonstrated technique and concluded that it was a highly specialized, non-generic bypass that only detected a small number of previously known, minor vulnerabilities that could be found by other publicly available models, such as GPT-5.5, without any bypass.
Anthropic complied with the legally binding directive but disagrees that the discovery of a highly specialized potential security bypass is grounds for revoking a commercial model deployed to hundreds of millions of users. Anthropic considers the incident a misunderstanding and is working to restore access. Until the situation is resolved, new sessions in Claude products will launch with Opus 4.8 or the user-selected AI model, and existing sessions with Fable 5 will terminate with an error.
Source: opennet.ru
