The 2026 bombing of Shajareh Tayyebeh primary school in Iran killed 175-180 children, yet media fixated on whether Claude chatbot selected the target. The real culprit was Maven, a Pentagon targeting system that pulls satellite imagery, signals intelligence, and sensor data to automate kill chains. This obsession with AI chatbots represents a dangerous distraction from the actual technologies causing harm.

Eight years ago, Google employees protested Maven, forcing the company to abandon its Pentagon contract. Palantir Technologies then spent six years building Maven into military infrastructure. The Minab school had been misclassified as a military facility in a Defense Intelligence Agency database that hadn't been updated since its conversion to a school in 2016. A chatbot didn't kill those children—outdated databases and automated systems did.

This "AI psychosis" affects critics and boosters alike, organizing attention around language models while ignoring older, more mature targeting technologies. The obsession with AI safety terminology like "hallucination" and "alignment" creates blind spots when discussing systems like Maven. The bureaucratic question isn't about Claude—it's about how Palantir's Maven Smart System compressed the kill chain from detection to strike, making outdated information lethally fast. Understanding this requires looking beyond charismatic technologies to the infrastructure actually causing harm.