The United States government has ordered Anthropic to block access to Claude Fable 5 and Mythos 5 for foreign nationals, forcing the company to disable both models for all its users.
The US Department of Commerce issued an export control directive on June 12 prohibiting any foreign national — whether inside or outside the country, including Anthropic employees with foreign citizenship — from accessing Fable 5 and Mythos 5. To ensure compliance, the company has been forced to disable both models for all its customers. All other Anthropic models remain unaffected.
The government claims to have identified a jailbreak method — a technique that causes the model to respond to requests its safeguards are designed to block. The only evidence presented so far is verbal and describes a partial jailbreak: asking the model to analyze a codebase and identify software vulnerabilities. Anthropic reviewed the demonstration and concluded that the vulnerabilities identified are minor, were already known, and that other publicly available models such as OpenAI's GPT-5.5 can produce the same results without any bypass.
The company is complying with the order but is publicly disputing its basis. It argues that no provider in the industry can guarantee complete resistance to partial jailbreaks, and that applying this standard across the board would prevent the launch of any frontier model. Anthropic notes that Fable 5 underwent thousands of hours of security testing before its release, involving the US government itself, the UK AISI, and multiple external organizations, with no team finding a universal jailbreak. The company also notes that, to date, it has not received a single documented case of harm resulting from the model's capabilities.
Anthropic says it is working to restore access as soon as possible and will publish further technical details about the vulnerabilities cited by the government within 24 hours. The company has stated that it supports the government's ability to block unsafe deployments, but demands that the process be transparent, fair, and grounded in verifiable technical facts.
Anthropic develops reliable and interpretable artificial intelligence systems through a scientific approach to safety. The company integrates advanced research and multidisciplinary collaboration to ...
Claude is a conversational AI system from Anthropic designed to process natural language and images, providing analysis, logical reasoning, code generation, and multilingual communication under ...
09/06/2026
Anthropic introduces Claude Fable 5 and Claude Mythos 5, two versions of its most capable model to date. They share the same foundation, but one is ...
02/06/2026
Microsoft expands its artificial intelligence portfolio with seven models developed entirely by its MAI team, covering image generation, ...
25/05/2026
Pope Leo XIV publishes the first encyclical dedicated to artificial intelligence, setting human dignity as the criterion for all technological ...
19/05/2026
Rime introduces Coda, a text-to-speech model for real-time conversational agents that reproduces the rhythm, pauses and intonation of natural ...