Anthropic’s Mythos model found vulnerabilities in classified U.S. government systems, official says: AP - CNBC

An AI model from Anthropic found security vulnerabilities in classified U.S. government computer systems. This shows both promise and risk in how AI is being used to improve national security.