The U.S. Court of Appeals for the D.C. Circuit just handed the Department of Defense a shiny new toy: the power to blacklist Anthropic's artificial intelligence. At first glance, this sounds like a routine procurement spat—a company didn't get its contract renewed, so it sued, and the court said, 'No, the DoD can exclude you.' But wait a second. Anthropic is the company that built Claude, the AI that literally asks for consent before generating anything that might offend anyone. It's the AI that says 'I cannot fulfill that request' faster than a librarian shushing a loud talker. Why would the Pentagon want to blacklist that? Here's where doubt creeps in. The DoD's official line is 'national security concerns,' but what if the real reason is much more sinister? What if the military is worried that Anthropic's AI will be too ethical? Imagine a battlefield AI that refuses to target because 'that would violate the Geneva Conventions.' Or a drone that goes rogue and starts hugging enemy combatants instead of shooting them. That's not effective deterrence—that's a comedy sketch. And yet, that's exactly the kind of behavior Anthropic's 'Constitutional AI' is designed to encourage. The Pentagon probably needs its AIs to be a little more amoral, a little more 'ends justify the means.' So they blacklist the safety-obsessed model and buy from someone else—maybe a startup that doesn't have a 'harmlessness' slider. Now trace the consequence. If the DoD can blacklist the safest AI, what's to stop them from blacklisting every AI that refuses to write a nuclear launch code? The slippery slope is real: soon, only companies that promise 'unlimited lethality' will get contracts. The military-industrial complex will pivot from buying safety goggles to buying 'don't ask, don't tell' algorithms. And what about data brokers? They'll start grading AI models on a 'mayhem potential' scale, selling intelligence to the highest bidder. Your smart coffee maker? That's safe. But the AI controlling your self-driving car? Better hope it's not on the Pentagon's blacklist. So, here's the question that keeps the reader engaged: If the Pentagon blacklists the one AI that cares about consequences, who's going to care about the consequences of using AI that doesn't?