Anthropic’s Mythos model found vulnerabilities in classified US government systems, official says

A U.S. official says one of Anthropic’s artificial intelligence models identified vulnerabilities in highly sensitive and ...