Anthropic disables its top AI models after US export control order

Anthropic has shut down access to its most powerful AI models, Claude Fable 5 and Claude Mythos 5, for all customers after the US government issued an export control directive citing national security concerns. The company said it received the order on Thursday evening and complied immediately, while publicly disputing the rationale behind the action.

The Directive

In a statement published on its website, Anthropic said the US government issued a directive requiring the company to suspend all access to Fable 5 and Mythos 5 by any foreign national, "whether inside or outside the United States, including foreign national Anthropic employees." Because the company cannot instantly distinguish between domestic and foreign users across its platforms, it disabled both models for all customers.anthropic

Axios reported that Commerce Secretary Howard Lutnick sent a letter to Anthropic CEO Dario Amodei stating that the models would be subject to export controls. Access to all other Anthropic models remains unaffected.9to5mac+1

The Jailbreak Dispute

The government's concern centers on a reported method of bypassing Fable 5's safety classifiers. Anthropic said it received only "verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws."anthropic

The company pushed back forcefully, stating it reviewed the technique in question and found it produced results "widely available from other models," including OpenAI's GPT-5.5. Anthropic added that it has not received disclosure of any jailbreak that "led to a harmful result" and called the government's action a "misunderstanding."anthropic

The episode follows a claim by a researcher known as "Pliny the Liberator," who said days earlier that he had bypassed Fable 5's safety classifiers and published the model's system prompt on GitHub.pasqualepillitteri

Broader Implications

Fable 5 had launched just days earlier on June 9 as the public-facing version of Anthropic's Mythos-class model, with new safeguards blocking responses in cybersecurity and biology. Mythos 5, the unrestricted twin, was limited to vetted government cyber defenders.thehackernews+1

Anthropic warned that if the standard applied to its models were used industry-wide, "it would essentially halt all new model deployments for all frontier model providers." The company said it is working to restore access and pledged to share more technical details within 24 hours.anthropic