A U.S. missile strike that destroyed a girls' school in Iran last month, causing significant casualties, is now suspected to be the result of a critical failure in an artificial intelligence targeting system. According to a forthcoming Pentagon report, preliminary findings indicate the school was misidentified as a military objective.

The strike, which occurred on February 28, was part of the broader 'Epic Rage' operation. Targets for that campaign were selected using the National Geospatial-Intelligence Agency's Maven Smart System, a platform that fuses surveillance and intelligence data for analysis. In a significant shift, Maven had recently been integrated with Anthropic's Claude language model to accelerate data processing.

Officials familiar with the investigation suggest the error may stem from the rapid expansion of the target list in the weeks preceding the operation. Hundreds of new potential sites were added, overwhelming the vetting process. The school's location was reportedly flagged because of its proximity to a suspected military complex, a connection that appears to have been incorrectly validated by the AI-augmented system.

While Defense Secretary Pete Heggseth has previously stated U.S. forces 'do not strike civilian targets,' the pending report is expected to formally acknowledge American responsibility for the incident. The investigation aims to determine whether the fault lies with flawed underlying intelligence, the AI's interpretation, or human oversight of the automated recommendations. The event has ignited intense internal debate over the growing reliance on machine-speed targeting in complex combat environments.

Source: Lenta.RU