Anthropic's Fable model was abruptly disabled by U.S. government order citing national security concerns over a jailbreak technique, exposing a core tension between the company's safety messaging and its economic need to release powerful models.

What happened

The U.S. government issued an export control directive suspending all access to Fable 5 and Mythos 5 for foreign nationals, both inside and outside the United States, after becoming aware of a method to bypass the model's safety guardrails. Anthropic received the directive at 5:21pm ET and disabled both models for all customers to comply. The jailbreak in question was used to identify a small number of previously known, minor vulnerabilities that other publicly available models can also discover.

Why it matters

Anthropic faces a structural economic problem — frontier AI labs lose billions building leading-edge models that are quickly distilled and commoditized by open-source alternatives, so they have strong incentive to move closer to users and lock in customers by becoming indispensable software rather than commodity inputs. This economic pressure to release capable models directly conflicts with the government's view that those same models pose national security risks, making friction between Anthropic and regulators inevitable as models grow more powerful.

What to watch

Senior Anthropic staff are currently in Washington D.C. seeking to resolve the dispute, with White House officials characterizing the company's position as dismissive of legitimate national security concerns. The company disputes whether the jailbreak found is universal or represents a fundamental flaw, and notes that Amazon — both an investor and major inference provider — reported the specific technique.