Use of AI in Iran elementary school massacre did not breach Anthropic's 'red lines': CEO

Anthropic’s CEO says the US military’s AI-assisted attack on Minab Elementary School did not breach the company’s red lines, though details of model use remain unclear. Dario Amodei asserted that the target was a human decision, not a machine choice, while acknowledging uncertainty about how the models were employed. The Feb. 28, 2026 strike, using Anthropic’s Palantir-enabled systems, killed more than 170 students and teachers at Minab, with another attack on a sports hall in Lamerd City killing 24 civilians. The broader conflict spurred a 40-day campaign that Iran criticized as aggression, with the Strait of Hormuz temporarily closed and ongoing regional hostilities.