Pentagon found AI flaws led to strike killing 123 Iranian school children

Military investigators determined that bad intelligence and algorithmic failures caused a missile strike on an Iranian school that killed 123 children. In a separate incident, the Pentagon nearly boarded a Chinese ship after AI systems produced false information about nuclear components onboard. Both cases revealed the military's heavy reliance on AI systems without adequate human verification or safeguards.

These aren't abstract concerns about AI bias or algorithmic fairness. They're concrete operational failures with lethal outcomes. The Pentagon deployed AI to make targeting decisions in combat without building in checkpoints to catch AI hallucinations or data errors. The fact that this came to light only through investigation — not through a disclosure or corrective action before the strike — underscores how the military normalized AI recommendations without systematic oversight.