TL;DR
A Pentagon investigation found that AI-assisted targeting, outdated imagery, and staffing cuts combined to kill at least 123 children when US Tomahawk missiles struck an Iranian elementary school in February 2026.
What happened
- February 28, 2026: Two Tomahawk missiles struck Shajarah Tayyebeh Elementary School in Minab, Iran, during an active US air campaign.
- 123 children killed per Bloomberg's account; a UN-backed fact-finding mission put the total civilian death toll at at least 178, including women and children.
- The school occupied a site that had previously been part of a military compound, but had been physically separated by walls and was operating openly as a school at the time of the strike.
- US Central Command personnel relied heavily on Palantir's Maven Smart System, an AI-enabled intelligence platform, during the targeting process.
- The US had struck more than 1,000 targets in the first 24 hours of its Iran campaign, compressing the time available for target verification.
Why it matters
- Pentagon investigators found personnel expected the AI system to flag stale or contradictory data rather than treating those gaps as triggers for human review, a critical doctrinal failure.
- Cuts to civilian-protection personnel created additional gaps precisely when operational tempo was highest, amplifying the risk of AI-assisted errors.
- Palantir denied its software was at fault for underlying intelligence failures, but developed new tools post-strike to re-examine intelligence and flag inconsistencies, implicitly acknowledging a process gap.
- The incident is now a landmark case for AI accountability in military targeting: it shows that automation can launder bad inputs rather than catch them when human oversight is degraded.
- International condemnation is mounting, with Italian PM Giorgia Meloni calling the strike a "massacre" and demanding accountability, signaling diplomatic costs beyond the battlefield.
What to watch next
- Whether the Pentagon formalizes new rules of engagement for AI-assisted targeting, including mandatory human verification steps when AI flags are absent.
- How Palantir's Maven Smart System is audited or restricted across other active theaters, and whether the Defense Department's broader AI acquisition strategy is revised.
- The UN fact-finding mission's final report, which could trigger international legal proceedings and set precedent for state liability when AI tools contribute to civilian casualties.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.