One of the more interesting AI stories right now has nothing to do with a new model release. Last year the Pentagon handed out $200M contracts to four AI companies — Anthropic (Claude), OpenAI, Google, and xAI. Same tools most of us use to write content, think through deals, run our businesses. Of those four, Anthropic's Claude ended up being the only one cleared for the Pentagon's classified networks. That's not a small thing. Then it got complicated. The Pentagon wanted contract language letting Claude be used for "all lawful purposes." Anthropic agreed to support the military broadly. But they held two lines: No fully autonomous weapons making lethal decisions without a human involved. No AI-powered mass surveillance of American citizens. Their CEO said they "cannot in good conscience" remove those safeguards. The Pentagon called them a supply chain risk. Contracts are being phased out. I'm not here to say who's right. What got my attention is simpler. we're watching an AI company walk away from hundreds of millions of dollars because of where they drew their ethical lines. And it raises a question worth thinking about for anyone building on these tools: When you pick your AI stack, how much do the values behind the company matter to you — not just how smart or capable the model is? What's your take?