Anthropic Crackdown: Amazon Uncovers AI Vulnerability

A dramatic crackdown on the AI company Anthropic was triggered by Amazon researchers. They discovered a "jailbreak" in Anthropic's new Fable 5 AI model. Fable 5 was released as a "safe" version of Anthropic's Mythos AI, which the company said had dangerous superhuman hacking skills. Amazon's team found a way to bypass Fable 5's safety features, potentially allowing access to information for cyberattacks. Amazon CEO Andy Jassy then informed White House officials about the vulnerability during a pre-scheduled call. He was encouraged to speak with Treasury Secretary Scott Bessent, who was already leading the administration's response to Anthropic's Mythos model due to its potential threat to the global financial system. This led to the U.S. government imposing export controls on Anthropic, forcing the company to pull its most powerful AI models. This marks the first time the U.S. has explicitly limited the release of a frontier AI model. This event has sparked global debate about AI regulation and its potential impact on technology and international relations.