When a company sells powerful tools to the military, you'd expect crystal-clear rules about who controls them and when. Instead, the Pentagon flagged the AI maker Anthropic as a risk after a string of 'holy cow' moments. The biggest one: the CEO reportedly suggested that during a missile-defense emergency, access to the AI could be sorted out with a quick phone call to him. A defense official's blunt reaction — in a real crisis, 'I'm not going to call you.'
Here's why that's a big deal: it shows nobody clearly knows who's in charge of these systems in the moment that matters. After a real military raid, the company itself had to call a contractor just to ask whether its own AI had been used.
So how does it touch you? Life-and-death systems — missiles, military strikes — are being wired up to AI that even its own makers can't cleanly account for. When the stakes are that high, 'just call me' isn't an oversight plan.