AI-driven targeting system linked to deadly school strike raises war-crime questions

During the first day of hostilities with Iran, a U.S. airstrike hit an elementary school that had been converted from a former naval base in 2017, killing 123 children. The attack was carried out using Maven, an AI-driven platform developed by Palantir Technologies that aggregates and prioritizes battlefield intelligence. Operators entered data labeling the school as an Iranian military installation, prompting Maven to list it as a high-priority target.

The intelligence package failed to reflect the school's civilian use, leading to the tragic mistake. Pentagon officials said some Central Command staff expected Maven to detect stale or inconsistent data, while a Palantir spokesperson argued the company is not responsible for the underlying data or any intelligence flaws. International law experts highlight the challenge of pinpointing criminal responsibility when autonomous tools and human commanders share decision-making, and they question whether existing legal frameworks can address such cases.