Anthropic Disables AI Models: US Cites National Security

Anthropic has disabled its new AI models, Fable 5 and Mythos 5, after the U.S. government cited national security concerns. The Commerce Department ordered the suspension of foreign access to these models on Friday. Here's the thing: the government's directive came after a reported method of bypassing the models’ safety restrictions. Secretary of Commerce Howard Lutnick issued a letter to Anthropic's CEO, Dario Amodei, placing the models under export controls. This prohibits use by foreign nationals, even foreign-born Anthropic employees. What's interesting is that Anthropic disabled the models entirely to ensure compliance. The company stated its understanding is that the government became aware of a technique for "jailbreaking" Fable 5. This involves circumventing the model’s built-in safety guardrails. Anthropic disputes the severity of this finding, saying similar capabilities are already in other public models and are used by cybersecurity professionals for defensive purposes. The company also acknowledged that perfect jailbreak resistance isn't achievable for any model provider. The bottom line: This action highlights growing concerns about the security and control of advanced AI technologies.