Pentagon preliminary investigation finds US responsible for Minab school bombing — outdated targeting data identified as cause

The Investigation Timeline

The investigation into the Minab school bombing unfolded in two distinct phases: independent media forensics (early March) and the US military’s own preliminary inquiry (reported March 6-11). Both converged on the same conclusion.

Independent Forensic Investigations (March 1-8)

Before the Pentagon acknowledged any investigation, independent organizations had already established the evidentiary foundation:

• Bellingcat (March 8): Conducted geolocation analysis of impact site imagery and debris field patterns. Their findings “appear to contradict” the claim by President Trump (made the previous day) that Iran was responsible. Bellingcat’s analysis confirmed the strike trajectory and weapon characteristics consistent with a sea-launched Tomahawk cruise missile.

• BBC Verify (March 6): Confirmed the weapon type as a US Tomahawk through independent fragment analysis and blast pattern assessment. The BBC’s verification unit operates separately from its editorial division, applying forensic methodology to physical evidence.

• CBC News (March 8): Published a visual investigation concluding the school was “bombed as part of a precision airstrike against the adjacent military complex” — the Sayyid al-Shuhada IRGC naval base. The CBC analysis established that the school and the military facility were distinct structures, clearly differentiated on commercially available satellite imagery.

Reuters Exclusive (March 6)

Reuters published the first report indicating the US military’s own investigation pointed to likely US responsibility. The exclusive, citing sources familiar with the inquiry, established that internal Pentagon analysis had reached the same conclusion as the independent forensic investigations. This was published the day before President Trump blamed Iran on Air Force One.

CNN Independent Analysis (March 6)

CNN’s analysis, published the same day as the Reuters exclusive but conducted independently, concluded the US was likely responsible. The convergence of two separate investigations from different outlets, arriving at the same finding on the same day, reinforced the evidentiary consensus.

Pentagon Preliminary Findings (March 11)

By March 11, four major outlets reported on the preliminary findings of the US military’s investigation:

• The New York Times: The preliminary inquiry found the US at fault. The root cause identified was outdated targeting data — the information used in the strike planning process did not accurately reflect the current state of the target area. The school’s presence adjacent to the military facility was not properly accounted for in the targeting package.

• CNN: Reported that outdated intelligence about the nearby Sayyid al-Shuhada naval base led to the strike. The intelligence used to plan the attack did not accurately characterize the proximity and function of civilian structures near the military target.

• The Guardian: Confirmed the preliminary investigation determined Washington was responsible. The incident was characterized as a “targeting mistake” — language that acknowledged error while stopping short of attributing intent.

• The Washington Post: Reported the target list may have mistaken the school for a military site. Critically, the Post raised the question of whether AI-enabled targeting systems contributed to the error — the first major outlet to explicitly connect the school bombing to the AI targeting pipeline that would become the subject of a separate investigation.

The AI Targeting Question

The Washington Post’s reporting on AI-enabled targeting introduced a secondary dimension to the investigation. The question was no longer only about outdated data — it was about whether automated systems processed that data, and whether AI-assisted collateral damage estimation models failed to flag the school as a protected civilian site.

This thread connected directly to NBC News’s March 11 reporting that Claude — Anthropic’s AI system — was actively being used in Iran air attack planning, and to the Pentagon’s same-day confirmation of a formal investigation into AI’s role in the targeting chain.

The convergence was significant: the school was bombed because targeting data was wrong. AI systems may have processed that wrong data. Those AI systems may have included tools whose safety guardrails the Pentagon had demanded be removed. Each layer of the investigation peeled back to reveal a deeper systemic question.

What Remains Unknown

The preliminary findings are not final. Key questions remain open:

• Who generated the targeting data? Was the outdated information a product of human intelligence collection, signals intelligence, satellite imagery analysis, AI-processed data fusion, or some combination?
• What review process approved the strike? Collateral damage estimation is a multi-step process involving legal review, commander authorization, and (in current doctrine) AI-assisted analysis. Which steps failed?
• Was the school specifically identified in the targeting package? “Outdated data” could mean the school was misidentified as a military structure, or it could mean the school’s existence was simply absent from the data set used.
• What role did AI systems play? The formal investigation into AI’s role (announced March 11) is separate from the targeting investigation and may take months to complete.

Until the final report is released, the confidence level for the investigation’s conclusions remains LIKELY rather than VERIFIED. The preliminary findings from multiple independent sources all point in the same direction, but the formal conclusion has not been issued.