Pentagon: Palantir AI Overreliance Led to Strike Killing 123 Iranian Children

Back in February, two Tomahawk missiles hit Shajarah Tayyebeh Elementary School in the southern Iranian town of Minab on the opening day of the Iran war. The blasts killed more than 150 people, including at least 123 children.

Earlier reporting already pointed to outdated targeting data and raised questions about whether artificial intelligence had a role in the mix-up. And now, a Bloomberg investigation published Friday fills in more of the story, citing officials involved in an unreleased internal Pentagon review. Those officials said some Pentagon personnel knew within hours that the United States had hit the school, and they described a cascade of preventable failures that included an overreliance on an AI tool built by Palantir.

Too Much Trust in AI

According to the officials who spoke to Bloomberg, some personnel inside U.S. Central Command leaned too hard on the artificial intelligence inside the Maven Smart System, an AI-powered data integration and targeting platform developed by Palantir to accelerate intelligence analysis and decision-making. The Defense Department has reportedly made the software tool a cornerstone of military operations over the past year. Several former senior officers now work for Palantir, including some with high-level clearances at Centcom’s Tampa headquarters.

Officials said some users expected Maven to flag stale records or contradictions in the intelligence assembled for potential targets, though it is not clear why they thought the system would do that. The Minab site, which was cataloged as an Islamic Revolutionary Guard Corps facility due to outdated data, was fed into Maven with other candidates and came out as a recommended day-one target. Target-list work that once took hours was condensed into minutes.

A Palantir spokesperson told Bloomberg the company “is not responsible for the underlying data nor identifying intelligence deficiencies” and that there is no evidence its software was at fault. Two people familiar with Palantir’s Pentagon contracts said the government keeps primary responsibility for the quality of the information that’s fed into Maven. After the strike, Palantir added features that “re-review underlying intelligence to identify factors that would disqualify a target and flag inconsistencies and inaccuracies that human review may have missed,” a person familiar with the work said. That new functionality is said to have already caught some anomalies.

More than 120 House Democrats wrote to Defense Secretary Pete Hegseth in March asking what role Maven and other AI tools played in identifying the site. The Pentagon has not answered publicly, citing the ongoing investigation. Officials told Bloomberg the full report has been all but complete for months. A Pentagon spokesperson said the incident remains under investigation and declined further comment.

Outdated Imagery of a Military Site That Had Become a School

Overreliance on Maven was only one of three failures in the kill chain described to Bloomberg. The other pieces were bad intelligence and outdated satellite imagery.

U.S. databases had listed the Minab compound as a military site for years, but commercial satellite images showed something different. Construction of walls and separate entrances that cut the school off from the adjacent base appears to have been finished by 2017. A 2018 image shows brightly painted walls, a soccer pitch, assembly rows, and playground markings.

Bloomberg reported in June that one analyst spotted changes as early as 2019 and logged remarks in a system that was not connected to the primary military intelligence database used for targeting. Those notes never reached the people who built the strike package. Human intelligence from southern Iran was thin, but the school had its own website and appeared on Google Maps.

The site stayed labeled as an IRGC facility as the Trump administration demanded an overwhelming opening assault. More than 1,000 Iranian targets were hit in the first 24 hours, and according to Bloomberg’s sources, that pace compressed the time available for additional confirmations.

Staffing on civilian harm mitigation teams across the U.S. Department of Defense had also fallen by roughly 90% over recent years, shrinking to fewer than 20 people overall. Centcom’s group shrank from 10 people to one. No member of those teams reviewed the Minab site before the missiles were up in the air.

UN Investigators Call It a War Crime

A separate inquiry by the United Nations’ Independent International Fact-Finding Mission on Iran this week reached the conclusion that there were reasonable grounds to believe the United States committed the war crime of launching an indiscriminate attack. It found the U.S. “failed in its obligation to do everything feasible to verify” that the school was a military objective and that the failure “went beyond mere negligence.” The report said the United States “directed the strikes at the building of the school while being aware of a substantial risk of striking a civilian object and acting recklessly as regards the possibility that this would happen.”

Despite these reports, Washington has not publicly accepted responsibility. In March, President Trump went as far as to tell reporters, “No, in my opinion, based on what I’ve seen, that was done by Iran. We think it was done by Iran, because they’re very inaccurate, as you know, with their munitions.” In a more recent Fox News interview, Trump said, “I don’t think anybody’s going to ever be able to say what happened there.” A senior administration official told Bloomberg, “The United States does not target civilians,” and the Defense Department has declined to comment on the UN report.

A Familiar Pattern of Errors in Military and Law Enforcement Use of AI

Errors tied to overreliance on AI are becoming a recurring theme as the technology becomes more embedded into military and law enforcement work. In a separate episode reported this week, a special operations analyst used a chatbot that hallucinated that a Chinese-flagged ship in the Middle East was carrying nuclear weapon components bound for Iran. The U.S. military prepared to board the vessel and pulled back at the last minute.

Law enforcement’s combined use of facial recognition and AI tech has also produced its own terrible outcomes. In Florida, an automated facial recognition match led to the arrest of a man who had never been to the state and was 400 miles away at work, resulting in months in jail and the loss of his job and home. In another case, a man was arrested based on a match despite being over 300 miles away at the time of the alleged crime. Charges in both cases were eventually dropped, but those cases are part of a much longer list of false arrests nationally based on software scores.

While people in all fields occasionally overrely on AI as they learn the practicalities of what it can and cannot do, the stakes are vastly higher in military and law enforcement applications, where human misuse or accepting automated outputs without verifying the underlying facts can carry catastrophic, life-or-death consequences. With that in mind, it is perhaps how humans are using this technology, rather than the theoretical uncontrolled power of it, that is the more pressing issue at this time.