A primary school in southern Iran was struck by American forces on February 28, killing between 175 and 180 children. Within days, the public debate centered on whether Claude, Anthropic’s chatbot, had selected the target. According to Hacker News, that framing missed the point entirely. The real story is far more unsettling.
The targeting system behind Operation Epic Fury wasn’t a large language model. It was Maven, a military targeting infrastructure built by Palantir Technologies over six years. Maven pulls together satellite imagery, signals intelligence, and sensor data to move targets through every step from detection to strike order. The school in Minab had been misclassified as a military facility in a Defense Intelligence Agency database that hadn’t been updated since at least 2016.
People failed to update a database. A system built for speed made that failure lethal. No chatbot was involved.
The Distraction Machine
What stands out here isn’t the technology failure itself: it’s how completely the public conversation missed it. Congress wrote to Defense Secretary Pete Hegseth about AI use in the strikes, but the questions were about Claude’s personality and self-preservation instincts. The New Yorker wondered whether a chatbot could be trusted to obey orders in combat.
Meanwhile, Maven, the system that actually enabled the strike, had become invisible. It had “sunk into the plumbing,” as the article puts it, becoming part of military infrastructure nobody questioned.
The article references scholar Morgan Ames’s concept of a “charisma machine”: certain technologies draw attention, resources, and attribution toward themselves and away from everything else. LLMs might be the most powerful example of this phenomenon ever. The vocabulary of “AI safety,” “alignment,” and “hallucination” has become so dominant that when something goes wrong with AI, those are the only terms people can reach for, even when they don’t apply.
Maven’s Quiet Rise
Here’s the history that matters. In 2018, over 4,000 Google employees signed a letter opposing Maven. Engineers quit. Google abandoned the contract. Then Palantir picked it up, and the protests stopped.
Maven became part of a longer military trend called “kill chain compression”: making the steps between detecting a target and destroying it shorter and faster. This thinking traces back to Obama’s second term and the “third offset strategy,” a bet that technological speed could compensate for strategic weaknesses.
Every generation of military technology has been sold on the promise of making kill chains shorter. Palantir’s Maven Smart System is the latest iteration. And because it’s infrastructure rather than a flashy chatbot, it operates with almost zero public scrutiny.
Why This Matters for the AI Industry
This story carries three important lessons:
- LLMs absorb all the oxygen. The entire AI policy conversation has narrowed to language models. Computer vision systems, targeting infrastructure, predictive analytics, and database-driven decision systems get a fraction of the regulatory attention, despite having far more direct consequences.
- The real risks are boring. An outdated database entry. A classification error nobody caught. A system optimized for speed that doesn’t pause for verification. These aren’t dramatic AI failure modes. They’re mundane ones. And they’re deadly.
- Corporate accountability shifts with attention. Google faced massive internal resistance over Maven in 2018. Palantir built the same system into full operational capability with barely a whisper of protest. When the public fixates on chatbots, defense contractors building targeting systems operate in relative silence.
For AI practitioners and policymakers, the takeaway is clear: the systems that need the most oversight aren’t necessarily the ones generating the most headlines. Regulation focused exclusively on foundation models and LLMs leaves enormous gaps in military AI, automated decision systems, and the databases that feed them.
The full analysis, which traces kill chain doctrine from 18th-century French artillery reforms to Palantir’s current contracts, is worth reading at the original source.