Killer AI is already here—we’re using it in Iran

Polls show most Americans fear AI could kill humanity, yet few know the US military already uses AI for lethal targeting in the war on Iran. Only 8 percent of Americans have heard "a lot" about this. According to reporting, the Pentagon's Maven Smart System, built by Palantir, uses Anthropic's Claude AI—the same tool people use for homework and emails—to semi-autonomously rank targets and draft legal justifications for strikes, with accuracy reportedly around 60 percent in good conditions and 30 percent in bad weather. Defense Secretary Pete Hegseth insists humans make lethal decisions, but critics argue human oversight is insufficient to prevent atrocities, citing the February US missile strike on a girls' school in Minab, Iran, that killed over 150 people and was later called a crime against humanity by UN experts. In the Iran war, the Pentagon struck more than 3,000 targets in the first week and 13,000 by the ceasefire 38 days later. AI targeting systems have also been used by Israel in Gaza. As AI companies enter the battlefield, experts warn the scale of atrocities is rising, and because the Department of Defense places few checks on tech companies—relying on self-regulation—corporate terms of service now shape how AI is used in war more than any law or executive guideline.

(Source：Mother Jones)