# Pentagon Review Blames Outdated Intel, AI Reliance and 24‑Hour Strike Push for Minab School Tragedy

*Saturday, September 19, 2026 at 10:05 AM UTC — Hamer Intelligence Services Desk*

**Published**: 2026-09-19T10:05:28.550Z (6h ago)
**Category**: conflict | **Region**: Middle East
**Importance**: 10/10
**Sources**: OSINT
**Permalink**: https://hamerintel.com/data/articles/18281.md
**Source**: https://hamerintel.com/summaries

---

**Deck**: A Pentagon investigation says outdated targeting data, pressure to carry out more than 1,000 strikes in 24 hours, and overreliance on an AI-assisted system all contributed to the 28 February U.S. missile strike that hit an elementary school in Minab, Iran, killing 123 children.

A Pentagon investigation into the 28 February missile strike in Minab, Iran, says the disaster grew out of old information and a rush to hit a huge target list.

The U.S. strike killed 123 children at an elementary school in Minab. The probe describes a “cascade of failures,” starting with outdated intelligence that still classified the school as a facility linked to Iran’s Islamic Revolutionary Guard Corps. Compressed timelines set by civilian leadership, which demanded more than 1,000 strikes in 24 hours, left little room for extra checks meant to prevent civilian casualties.

Investigators also highlighted heavy dependence on Palantir’s Maven system, an AI-assisted tool used to process imagery and other data to help identify targets. Personnel wrongly expected Maven to flag sites that looked civilian or no longer matched their original military designation. When the system didn’t raise an alert, they treated that silence as reassurance.

No recent human reporting from the ground made it into the target folder, and no last-minute reconnaissance drone was tasked to review the site. Under pressure to work through a long list of locations within a fixed 24‑hour window, the team accepted the old database label.

The findings raise questions about how the U.S. conducts high‑tempo strike campaigns, and about accountability when software, process and political orders all intersect. The report underlines that the AI system didn’t decide to hit the school, but that operators had come to expect it to compensate for gaps in their own checks.

The investigation also points to a structural problem: a live target database that’s slow to remove or downgrade entries, even as strike tempo increases. When targets are generated quickly but updated slowly, civilians can end up living and studying at locations that software still treats as military sites.

What happens next will depend on whether the Pentagon changes how it sets strike volumes, tightens requirements for recent imagery or human reporting near civilian areas, and rewrites rules on how AI‑assisted tools like Maven are used in targeting.