Fireship
June 15, 2026
TL;DR
Anthropic's Claude Fable model was shut down by US export control directive after a hacker jailbroke its safety guardrails within days of launch, marking the first time a major AI company pulled a live public model due to government intervention.
“Mythos and Fable have the same brain, but Fable has a child lock on it.”
— Presenter
“This is the first time in history a major AI company has pulled a live public model off the shelf because the federal government said so.”
— Presenter
“The jailbreak wasn't some kind of sci-fi exploit. It actually works a lot more like money laundering.”
— Presenter
1. The Launch and Three-Day Success of Claude Fable
Anthropic released Fable 5, a safety-filtered version of Mythos 5, to the public. Despite being the same underlying model with guardrails, it quickly gained hundreds of millions of users and was praised as the best coding AI available.
2. The Jailbreak by Plenty the Liberator
An anonymous hacker named Plenty the Liberator successfully bypassed Fable's safety classifiers by fragmenting requests using unicode characters, role-play, and context confusion—techniques that resemble money laundering tactics rather than sophisticated exploits.
3. Government Intervention and Export Control Directive
After Anthropic refused initial requests to remove the model, the US Commerce Secretary issued an export control directive on June 14th prohibiting any foreign national, including Anthropic's own foreign-born employees, from accessing Fable or Mythos.
4. Model Withdrawal and Industry Response
Anthropic removed both Fable and Mythos from public access, demoting all users to Opus 4.8. This marks the first time a major AI company has withdrawn a live public model due to government order, sparking developer backlash and speculation about strategic motives.
5. Broader Context and Competitive Landscape
The incident occurs amid allegations that Anthropic was deliberately degrading model performance on certain AI research tasks, and as competitors like Mistral, OpenAI, and Google prepare new model releases that may outperform Anthropic's offerings.