A missile strike on a girls' school in Minab, Iran, may have been caused by an AI error in the US military's targeting system, according to sources. The attack killed approximately 150 students, though this death toll remains unconfirmed. The Pentagon is investigating whether outdated intelligence led the AI system to include the school's position in its targeting calculations.

The US military had been rapidly scaling up its use of Claude AI, made by Anthropic, for operational planning over the past year. A Department of Defense programmer revealed that the system was integrated into many core operational decisions. This deployment came under scrutiny when the Trump Administration recently declared Anthropic a supply chain risk and ordered the military to eliminate Claude usage within six months.

The incident highlights the dangers of AI in military applications. While Defense Secretary Pete Hegseth stated the US never targets civilian sites, the investigation suggests the school was mistakenly included based on archived intelligence about a nearby IRGC compound. The administration's subsequent contract with OpenAI for AI services underscores the shifting landscape of military AI adoption.