The Pentagon's internal review, described to Bloomberg, links Palantir's Maven AI targeting tool to a Feb. 28 Tomahawk strike on Shajarah Tayyebeh Elementary School in Minab that killed more than 150 people, including about 120–123 children. Maven recommended the site because U.S. databases still listed it as an IRGC facility, even though satellite imagery and open-source records showed it was a school. The review cites preventable failures across the targeting chain, major cuts to civilian-harm mitigation teams, and poor integration of analyst observations. The incident raises urgent accountability questions as AI accelerates battlefield decision-making.
Pentagon Review Ties Palantir AI Tool To Tomahawk Strike That Killed About 120 Children In Iran

On Feb. 28, two Tomahawk missiles struck the Shajarah Tayyebeh Elementary School in Minab in southern Iran on the opening day of the U.S.–Iran war. Investigations by Bloomberg and reporting from the BBC conclude that more than 150 people were killed, including roughly 120–123 children. An internal Pentagon review, described to Bloomberg by officials involved in that process, identified a cascade of preventable failures in the kill chain that led to the school being selected as a target.
Key technology in the process
Palantir's Maven Smart System, an AI-driven platform operated under a Pentagon contract, fuses more than 150 data sources — satellite imagery, drone feeds, signals intelligence and other inputs — to produce targeting recommendations. During the initial phase of the war, when officials pushed for strikes on hundreds of sites within hours, Maven compressed days or hours of target-list work into minutes and returned the Minab compound as a recommended day-one target because U.S. military databases still labeled it an IRGC facility. The system performed as designed; the critical failure was that no process required it to validate whether the underlying classification reflected on-the-ground reality.
Disconnects between data and reality
Although U.S. databases retained an IRGC classification for years, satellite imagery and open-source information indicated the site had been converted to a school. By 2017 imagery showed new walls and separate entrances isolating the compound from adjacent military sites; 2018 images displayed brightly painted walls, a soccer pitch and playground markings. An analyst reportedly observed these changes as early as 2019 but logged the observations in a system that was not connected to the primary targeting database, so those notes never reached the team assembling the strike package. The school also maintained a public website and appeared on Google Maps, but that open-source evidence was not integrated into the targeting workflow, according to reporting.
“The Mission finds \"reasonable grounds to believe that the US committed the war crime of launching an indiscriminate attack resulting in loss of life or injury to civilians or damage to civilian objects.\"” – Independent International Fact-Finding Mission on Iran, as reported by the BBC.
Organizational and oversight failures
The internal review and related reporting say civilian-harm mitigation staff across the Department of Defense fell by about 90%, leaving fewer than 20 specialists; CENTCOM's dedicated mitigation team shrank from roughly 10 to a single analyst. The review found that no member of those reduced teams examined the Minab target before missiles were launched.
Responses and accountability
U.S. officials have issued denials and partial defenses. In March, President Donald Trump said he believed Iran was responsible, and Pentagon officials have emphasized that the U.S. does not target civilians. Palantir told Bloomberg it is "not responsible for the underlying data nor identifying intelligence deficiencies" and that no evidence shows its software itself caused the strike. Contract experts note that the U.S. government bears primary responsibility for data quality and target verification. After the strike, Palantir reportedly added Maven features intended to flag disqualifying indicators and surface inconsistencies that human review might miss.
More than 120 House Democrats have written to Defense Secretary Pete Hegseth requesting detailed answers about Maven’s role in selecting the Minab target. The Pentagon has not publicly released the internal review, citing an ongoing investigation. The episode raises urgent questions about who is accountable when AI compresses decisions that once required extensive human scrutiny: who is responsible for what the system does not know? At Minab, that question remains unresolved, and the human toll is stark — more than 120 children were killed.
Help us improve.






















