Anthropic’s Mythos model found vulnerabilities in classified US government systems, official says - Federal News Network

An AI model identified security flaws in U.S. government classified systems, raising important questions about AI safety and government cybersecurity. This affects national security and public trust in government IT systems.