Anthropic CEO Dario Amodei told Bloomberg on June 10 that his company’s Claude AI model did not cross Anthropic’s established “red lines” when it was used during the US Tomahawk cruise missile strike on a girls’ elementary school in Minab, Iran. The strike, which occurred on February 28, killed an estimated 120 to 168 people, predominantly children under the age of 12.
What happened in Minab
On February 28, the first day of a significant US military campaign against Iran, a Tomahawk cruise missile struck a girls’ elementary school in Minab. The casualty estimates range from 120 to 168, with the vast majority being children.
Claude AI was integrated with Palantir’s Maven system during the operation. Maven is a military platform designed for rapid target analysis and processing. AI helped sort through data to identify and evaluate targets during the campaign, and one of those targets turned out to be a school full of kids.
Amodei’s defense, delivered months after the strike, rested on a specific distinction. Human operators, he emphasized, retained ultimate decision-making authority over military actions. Claude assisted in the process. It did not pull the trigger.
