In a bizarre yet not unforeseen twist of events, the US government ordered Anthropic to immediately disable Claude Fable 5 and Claude Mythos 5 for all users worldwide. Anthropic received the directive at 5:21 pm ET and complied within hours, taking both models offline globally, not just for the foreign nationals the export control order nominally targeted. Access to Anthropic’s other models remains unaffected.
Fable 5 launched just three days before the shutdown order, positioned as Anthropic’s answer to commercial pressure: a version of the highly restricted Claude Mythos fitted with independent classifier systems blocking high-risk outputs in cybersecurity and biology. Anthropic argued those classifiers operated separately from the model itself, meaning a jailbreak of the conversational layer left the core safety protections intact.
The government’s underlying concern, according to Anthropic, centers on a claimed narrow, non-universal jailbreak of Fable 5 that involves prompting the model to read a codebase and identify software flaws. Anthropic disputes that this capability justifies a global recall, noting that the same level of capability exists in other publicly available models including OpenAI’s GPT-5.5, and that cybersecurity professionals routinely use similar techniques for defensive purposes. The government provided only verbal evidence of the jailbreak rather than technical documentation.
Anthropic made clear it disagrees with the decision.
We are complying with the government’s legal directive and are removing access to Fable 5 and Mythos 5 for all users. However, we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.
The shutdown carries a painful irony for Anthropic. The company spent months publicly warning that Claude Mythos was too dangerous for general release, restricting it to roughly 50 vetted organizations through Project Glasswing.
Just a few days ago, Anthropic CEO Dario Amodei published a sweeping policy framework just days earlier calling for FAA-style government regulation of frontier AI models and urging democracies to take AI risks seriously. That essay may have landed differently than intended.
Those warnings, intended to demonstrate safety leadership, now appear to have directly attracted the government scrutiny that forced a global shutdown of Fable 5 just days after launch.
