AI-enabled war on Iran
2026-03-22 - 21:41
Palantir’s Maven Smart System is the C2 (Command and Control) platform at the heart of the war on Iran. The war on Iran is different from all the wars fought before it – because this is the first large-scale high-intensity conflict in which AI (Artificial Intelligence) is not just a tool but the operational engine driving the entire joint targeting cycle. Maven ingested 179 live multi-domain data feeds — satellite imagery, drone ISR (Intelligence, Surveillance, and Reconnaissance) video, radar tracks, SIGINT (signals intelligence), and geolocation data — and fused them in real time into a COP (Common Operating Picture) fed simultaneously to commanders across all echelons. Claude, Anthropic’s large language model integrated into Maven, then processes this operational picture to identify, classify, and score targets — assessing each by entity type, threat level, and time-sensitivity. In the first 24 hours of Operation Epic Fury, Claude surfaced over a thousand actionable targets; by day thirteen, American forces had struck over 15,000 targets across Iran. Maven automatically prioritises the resulting target queue. A human commander then reviews each recommendation and authorises the strike. Maven matches the approved target to a strike asset — aircraft, precision-guided munition, loitering munition, or cyber effect — and the strike is executed. Post-strike, AI conducts Battle Damage Assessment (BDA) and reattack decisions re-enter the queue within minutes. What once required a targeting cell of roughly 2,000 personnel during Operation Iraqi Freedom (2003) is now being executed by a cell of 20, at a rate of 80 targets per officer per hour (Wikipedia). That is the complete AI-enabled kill-chain – from sensor to shooter, in near-real time. Maven’s legal legitimacy under IHL (International Humanitarian Law) and the Law of Armed Conflict (LOAC) rests on the assurance that a human being, not a machine, bears command responsibility for lethal force. Consider the arithmetic. At 80 targets per officer per hour, a commander is spending roughly 45 seconds on each authorisation — reviewing an AI-generated recommendation, assessing proportionality, checking compliance with international humanitarian law, and approving a strike. Whether 45 seconds constitutes meaningful human control, or whether it has become a procedural rubber stamp on decisions the machine has already effectively made, is the question that lawyers, ethicists, and military commanders have not yet resolved. Operation Epic Fury has not answered it. It has simply made it impossible to ignore. Maven is now fielded across all eleven US Geographic Combatant Commands (GCCs) and has been acquired by NATO’s Allied Command Transformation. What began as a DARPA-adjacent Pentagon experiment in drone video analysis in 2017 has, in less than a decade, become the permanent warfighting architecture of the Western joint force. Iran is not the destination. Iran is the demonstration. Every General Staff that watched Operation Epic Fury is now asking one question: how quickly can we stand up our own Maven? The age of AI-enabled multi-domain operations has not arrived. It has been normalised. —The writer is a journalist and political analyst.