Former national cyber director: Anthropic's 'Mythos' AI can hack nearly anything and we aren't ready - Fortune
Anthropic's most powerful model yet is less a product launch than a stress test—one that exposes dangerous gaps in how the U.S. protects its critical systems.