Pentagon Review Blames AI Overreliance for Deadly Minab School Strike

3 min read
Source: Gizmodo
Pentagon Review Blames AI Overreliance for Deadly Minab School Strike
Photo: Gizmodo
TL;DR

A Bloomberg investigation reveals that U.S. military personnel relied too heavily on Palantir’s Maven AI system, which failed to flag outdated data identifying a school as a military target. This error, combined with thin staffing and rushed timelines, led to a strike killing over 150 people, including 123 children. While the UN calls it a war crime, legal experts argue the evidence for criminal recklessness is weak.

Key points

  • Bloomberg reports that Centcom personnel over-relied on Palantir’s Maven system, which did not flag that the Minab target was a school rather than an IRGC facility.
  • The strike on February 28, 2026, killed at least 156 people, including 123 children, due to outdated satellite imagery and a 90% reduction in civilian harm mitigation staff.
  • The UN’s Independent International Fact-Finding Mission concluded there are reasonable grounds to believe the U.S. committed a war crime by launching an indiscriminate attack.
  • Palantir stated its software is not responsible for underlying data quality, while the Pentagon has not publicly accepted responsibility for the strike.
  • Legal scholars argue the UN’s war crime conclusion is unconvincing because it fails to prove specific personnel acted with subjective recklessness rather than negligence.

Background

This incident follows a pattern of AI-related errors in military and law enforcement contexts, including a recent case where a chatbot hallucinated a nuclear threat on a ship in the Middle East. It also occurs amid broader tensions, including Iranian strikes on U.S. bases in the Middle East and ongoing diplomatic efforts to restart negotiations during the UN General Assembly.

How outlets are covering it

Gizmodo and Bloomberg emphasize the technical failure of the Maven system and the catastrophic impact of reduced staffing on civilian harm mitigation. The UN and CNBC highlight the legal and diplomatic fallout, noting the U.S. rejection of the war crime findings. In contrast, Opinio Juris argues that while the attack violated international humanitarian law, the evidence does not support the specific legal definition of a war crime based on reckless intent, suggesting the failure was one of negligence rather than criminal recklessness.

Why it matters

The incident highlights the risks of integrating AI into high-stakes military targeting systems, where automated outputs may override human verification. It also underscores the consequences of budget cuts to civilian harm mitigation teams and the potential for diplomatic and legal repercussions when such errors result in mass civilian casualties.

What to watch

The Pentagon is expected to release the full internal review, though it has been delayed for months. Diplomatic talks between the U.S., Iran, and Gulf Arab nations may resume around the UN General Assembly. Palantir has added features to its software to re-review intelligence and flag inconsistencies, but the legal and political debate over responsibility for the strike continues.

Share this article

Want the full story? Read the original reporting

Read on Gizmodo