About This Article
Anthropic's Claude Mythos found thousands of code vulnerabilities in weeks — a Move 37 moment for computer security. But the deeper threat is what happens when that same paradigm-breaking capability targets human psychology: personal data as attack surface, persuasion techniques we haven't imagined, and a trust model built on assumptions that AI is about to falsify.