Pentagon Review Links AI Reliance to Deadly School Strike

A recent internal review by the Pentagon indicates that an overreliance on artificial intelligence tools contributed to a U.S. military strike that killed at least 123 children. The incident, which occurred in February when two Tomahawk missiles hit an elementary school in Minab, Iran, has raised serious questions about how military targeting decisions are made in the age of automated systems.

According to officials familiar with the unreleased report, personnel at U.S. Central Command relied heavily on the Maven Smart System, an AI platform developed by Palantir. The system is designed to accelerate intelligence analysis by integrating various data sources. However, the review found that this reliance masked critical errors in the underlying data, leading to a catastrophic failure in the targeting process.

AI systems masked human errors

The Maven system was expected to flag contradictions or outdated records in the intelligence provided to it. In the case of the Minab school, the site was incorrectly cataloged as a facility for the Islamic Revolutionary Guard Corps due to stale databases. Because the system did not flag these inconsistencies, the school was included in the list of recommended day-one targets. This automation condensed what used to be hours of manual verification into minutes, potentially bypassing critical checks.

Palantir has stated that it is not responsible for the quality of the data fed into its software. A spokesperson emphasized that the government retains primary responsibility for identifying intelligence deficiencies. Following the incident, the company added new features to re-review underlying intelligence and flag anomalies that human review might have missed, suggesting that the previous version of the tool lacked sufficient safeguards for such high-stakes decisions.

Outdated data fueled the mistake

The failure was not solely due to the AI tool. Commercial satellite imagery had shown that the site had been converted into a school by 2017, with visible features like a soccer pitch and painted walls. Despite this, U.S. databases continued to list the compound as a military site. An analyst had even noted the changes in 2019, but these remarks were logged in a system disconnected from the primary targeting database, meaning they never reached the decision-makers.

This disconnect highlights a broader issue in military intelligence workflows, where data silos can prevent critical information from influencing operational decisions. The combination of outdated imagery, fragmented intelligence sharing, and an AI system that did not adequately challenge the input data created a perfect storm for error. The Pentagon has maintained that the incident remains under investigation, with the full report reportedly completed but not yet released.

Questions remain on accountability

Over 120 House Democrats have written to the Defense Secretary asking for clarity on the role of AI in identifying the strike target. The Pentagon has cited the ongoing investigation to avoid public comments. This incident serves as a stark reminder of the trade-offs involved in integrating AI into lethal military operations. While the technology promises speed and efficiency, it can also obscure human oversight if not carefully managed and monitored.

As reported by GN technics/ai (en-US), the case underscores the need for robust verification processes in automated systems. The tragedy in Minab is a cautionary tale for the defense sector, highlighting that efficiency gains must never come at the cost of accuracy or ethical responsibility. The full implications of the review may reshape how AI is deployed in future military operations.