The morning of February 28, 2026 was another ordinary Saturday for Minab, Iran. Traffic flowed as normal, parents dropped their children off at school, and students began to practice volleyball. At around 10:00 AM IST (Iranian Standard Time), President Trump ordered the United States military to commence Operation Epic Fury. Airstrikes rained down on Minab. By the end of the hour, the Shajareh Tayyebeh (“Sacred Tree”) elementary school was reduced to rubble. A total of 168 people were killed, 120 being schoolchildren. 

As of May 1, 2026, the attack on Minab is the single deadliest event of the United States’ joint war with Israel against Iran. As the world demanded answers as to how such a horrific tragedy could happen at the hands of the United States, people began to point fingers at AI tech error. 

Given the U.S. military’s use of Gen-AI, the bombing of Shajareh Tayyebeh shows the dangers of integrating such a powerful technology in war. 

The beginning of Operation Epic Fury on February 28 involved air strikes against sites deemed to pose an “imminent threat.” These sites primarily consisted of locations along Iran’s Southern and Western borders, including “headquarters and technology sites” of the Islamic Revolutionary Guards Corps (IRGC), “missile and drone manufacturing sites,” and other general Iranian military infrastructure. 

Shajareh Tayyebeh had likely been classified as a target because the building itself was formerly a part of the nearby Sayyid al-Shuhada military complex. According to digital investigations by several news outlets, including Al Jazeera, The Guardian, and Reuters, the school has been separate from the military complex for at least a decade. 

In fact, it had been clearly walled off from the complex since at least 2016. This is evidenced by satellite imagery and the school’s official website (now removed), which showed decorated classrooms, worksheets, and children playing—all indicators that the school was a civilian institution. 

Independent journalism and research analyses also revealed that the U.S. military used Gen-AI technology to generate its targets, namely, the Department of Defense’s Maven Smart System. The Maven Smart System (or simply “Maven”) is a machine learning algorithm launched by the U.S. military used to integrate advanced data analytics and automatic target recognition into the DoD. 

Maven uses Palantir Technologies‘ data integration to analyze vast amounts of surveillance data, including satellite imagery and geolocation data. From 2024 to March 2026, Anthropic’s Claude AI served as its Large Language Model interface. The October 7 attacks in 2023 prompted the U.S. military to begin using Maven for combat purposes. Most recently, it was used to generate coordinates for the February 28 attacks, and independent researchers theorize that the system used outdated data to determine these targets.

In addition, it is possible that, over the course of Maven’s use and data training, the system developed a bias. According to Hanna et al. (2025), a rising concern with artificial intelligence and machine learning is that the tool’s output may be influenced by how it is used over time. Because Gen-AI is a predictive technology, the information it searches for may be more likely to confirm the user’s request. 

That is to say—in the case of the February 28 attacks—Maven was likely only trained to detect a location’s positive relevance to the IRGC/Iranian military, rather than to additionally check for up-to-date information that would disprove such relevance. For Shajareh Tayyebeh, Maven either ignored or was not fed any information indicating that the site no longer functions militarily. 

It is important to recognize that labeling the attack on Minab as an “AI error” overlooks the responsibility of those controlling the operation. Maven was integral to conducting the strikes, but the U.S. military’s use of an incomplete technology without proper oversight ultimately caused the deaths of hundreds of innocent Iranians, most of whom were children. 

Given that the October 7 attacks spurred the premature deployment of Maven for combat, it appears the U.S. military is desperate to have Maven as its “offset” technology against other countries. Writer and researcher Kevin Baker points out in his article: 

“Organizations that run on the formal procedure need someone inside the process to interpret rules, notice exceptions, [and] recognize when the categories no longer fit the case.” 

Leading up to February 28, there was no one who bothered to run a simple check on the target coordinates. One simple search would have quickly revealed that Shajareh Tayyebeh was a school. There are currently no policies in place that require human oversight of the use of Gen AI technology in the military, let alone in cases that could put civilian lives at risk. 

Even if it was unintentional, the bombing on Shajareh Tayyebeh is already being labeled as a war crime committed by the United States against the people of Iran, by organizations such as Human Rights Watch and countless international law experts. 

The beginning of the war in Iran already shows us that Gen-AI is powerful but prone to error. The U.S. government’s lack of initiative to place restrictions on its implementation in the military will pose future threats to whoever happens to be deemed—or rather, generated—the target enemy. Without pressure, imperfect technologies like Maven will continue to be used to make deadly decisions by militaries thousands of miles away from the people they impact.