On July 1, Anthropic released a lengthy article on Tuesday local time, detailing the entire process of the two AI models, Claude Fable 5 and Claude Mythos 5, from being subject to U.S. export controls and suspended access to ultimately resuming their launch. This announcement not only focuses on the restoration of access but, more notably, highlights Anthropic's efforts to transform the 'severity of AI model jailbreaking' into an industry-wide scoring system and to promote a new process requiring government pre-launch testing and safety collaboration before the release of cutting-edge models.
