US Military AI Targeting Platform Fed Stale Data Contributes to Strike on Iranian Elementary School (2026)

Content note: This entry describes a military strike that killed children.

Incident Summary

On February 28, 2026 — one hour into the opening day of the US-Israel military campaign against Iran — a US Tomahawk missile struck the Shajareh Tayyebeh elementary school in Minab, Iran, located approximately 100 yards from an IRGC naval complex.^[1][3] Sources report between 165 and 175 people killed, including more than 100 children — Iranian authorities reported 165 dead, while the Washington Post and Human Rights Watch cited at least 175.^[1][2][3] A preliminary US military investigation found that the strike resulted from reliance on outdated, human-curated target coordinates provided by the Defense Intelligence Agency that had not been updated since the school replaced a former military facility at the same location.^[1][2] Project Maven — the Pentagon’s AI-assisted targeting platform — processed this stale data and recommended the target for human approval. Maven is not a fully autonomous engagement system; a human commander authorized the strike. Former military officials have stated that “humans — not AI — are to blame” for the error, pointing to the stale human-curated data fed to the system.^[4]

Further reporting revealed that the Civilian Protection Center — the Pentagon unit responsible for reviewing targeting decisions to minimize civilian harm — had its workforce cut by approximately 90% prior to the strike, severely degrading the human oversight layer intended to catch targeting errors.^[4][5] The UN described the bombing as a “grave violation of humanitarian law.”^[3] The incident has prompted international calls for moratoriums on AI-assisted military targeting and accountability for the systemic failures — in data curation, human oversight, and target verification — that allowed stale coordinates to drive a lethal strike on a school.

Key Facts

• Casualties: Between 165 and 175 people killed (sources vary), including more than 100 children, at the Shajareh Tayyebeh elementary school in Minab, Iran^[1][2][3]
• Timing: The strike occurred one hour into the opening day of the US-Israel military campaign against Iran, at approximately 10:45am local time^[3]
• AI system: The Pentagon’s Project Maven AI-assisted targeting platform (human-in-the-loop, not fully autonomous) processed outdated, human-curated intelligence data that contributed to target misidentification; a human commander authorized the strike^[1][4]
• Root cause: Target coordinates from the Defense Intelligence Agency had not been updated since the school replaced a former military facility at the same location^[1][2]
• Proximity: The school was located approximately 100 yards from an IRGC naval complex^[1]
• Oversight failure: The Civilian Protection Center’s workforce had been cut approximately 90% prior to the strike^[4]
• International response: Human Rights Watch and Amnesty International launched formal investigations; the UN called the bombing a “grave violation of humanitarian law”^[2][3]
• Weapon system: Tomahawk cruise missile, used exclusively by US forces in this conflict^[3][4]

Threat Patterns Involved

Primary: Lethal Autonomous Weapon Systems — While Project Maven is an AI-assisted targeting system (not a fully autonomous engagement system), it processed outdated human-curated intelligence data that contributed to the misidentification of a school as a military target. Former military officials have stated that “humans — not AI — are to blame,” pointing to stale data curation rather than algorithmic failure.^[4] Nevertheless, the incident illustrates how AI-assisted targeting systems amplify the consequences of upstream data errors at machine speed and scale. The 90% reduction in Civilian Protection Center staff effectively hollowed out the human-in-the-loop safeguard designed to catch exactly this type of error, creating a situation closer to “autonomy by default” despite the nominally human-in-the-loop architecture.

Secondary: Overreliance & Automation Bias — The systemic reduction in human oversight capacity, combined with reliance on AI targeting recommendations based on stale data, reflects institutional overreliance on automated targeting systems and a failure to maintain adequate human judgment in lethal decision-making.

Significance

1. Severe AI-assisted targeting failure — With between 165 and 175 civilian casualties including more than 100 children, this incident represents a severe documented case of AI-assisted military targeting contributing to civilian harm, establishing a stark reference point for the consequences of stale data in AI targeting pipelines
2. Structural oversight degradation amplified AI risk — The 90% reduction in Civilian Protection Center staff prior to the strike illustrates that AI targeting systems become catastrophically more dangerous when human oversight infrastructure is simultaneously degraded — a pattern of structural negligence with implications for any domain where AI replaces rather than augments human judgment
3. Data quality as a single point of failure — The use of “stale intelligence” as input to the Maven targeting system illustrates that AI targeting accuracy is fundamentally constrained by data freshness and quality, and that no amount of algorithmic sophistication can compensate for outdated inputs in life-or-death decisions
4. International legal and humanitarian implications — The incident has intensified calls for international regulation of AI-assisted military targeting, with humanitarian organizations documenting the case as evidence for binding protocols on autonomous weapons systems