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

An AI model from Anthropic named Mythos has detected security flaws in classified U.S. government systems during a collaborative testing phase, according to a U.S. official. This discovery raises concerns about the potential risks artificial intelligence poses to national security and underscores the need for rigorous oversight and updated cybersecurity protocols. The incident highlights the dual-edged nature of AI advancements, which can both enhance and threaten critical infrastructure.