The US Government froze Anthropic's Claude Fable 5 on 12 June, two days after its release on 9 June, over fears that the model's automated hacking capabilities were escaping regulatory boundaries. Claude Fable 5 and Claude Mythos 5 are the two variations of Anthropic's most advanced AI tier; Mythos 5 is a non-public version restricted via Project Glasswing for government agencies, while Fable 5 is the public-facing version built on that same technology. Anthropic modified Fable 5 with built-in safety filters before release. The federal mandate affects foreign citizens inside the US, including researchers employed at Anthropic, and the company cut off access for all customers globally to ensure compliance. The US Government asked Anthropic to prevent all foreign nationals from accessing both models, citing a method of bypassing, or jailbreaking, the model. Anthropic claims it reviewed a demonstration of the bypass technique but maintains it only exposed minor, previously known security flaws. White House Adviser David Sacks wrote that officials issued the export control reluctantly, stating that CEO Dario Amodei refused to fix the jailbreak or de-deploy the model, and that the hope is Anthropic remediates the issue so the export control is lifted.