Anthropic's Mythos Model Found Vulnerabilities in Classified U.S. Government Systems, Official Says

My take: Look, hearing that Mythos found flaws in classified U.S. government systems in hours, not weeks, gives me goosebumps (the good and the bad kind). That Anthropic works with intelligence inside Project Glasswing and defensive cybersecurity makes total sense, that is not the story. What really makes me think is that this capability ALREADY exists: AI reviews insanely complex systems at a speed no human team can match, like a locksmith testing every door in a whole building before you finish your coffee. That hugely empowers whoever uses it to defend (and that is why I am glad it is Claude on the good side), but it also means a bad actor with a similar model is a concrete threat, not a movie plot. The question nobody in government wants to say out loud is what happens when this power is no longer only Anthropic's. Sure, that is my opinion.