Ukrainian forensic teams say they have documented what appears to be the first civilian deaths caused by a Russian drone using fully autonomous targeting.
According to Tom's Hardware, a Russian Molniya drone carrying an Nvidia Jetson Orin module struck a gas station in Zaporizhzhia last month, killing three civilians. The Jetson Orin is a compact computing module built for running AI models on devices in the field — the kind of hardware normally marketed for robotics and industrial machines, not weapons.
The New York Times, as reported by The Insider, described the drone as an experimental AI-controlled system, and framed the Zaporizhzhia attack as the first documented case of civilians killed by such a weapon.
Separate reporting aggregated by Bing News says circumstantial evidence has confirmed that Russia has deployed "autonomous lethal drones" in Ukraine — machines that select and strike targets without human instruction.
That last phrase is the heart of the story. Conventional attack drones still have a person in the loop, watching a video feed and deciding when to strike. A system that picks its own target removes that human judgment from the moment of killing. Arms-control researchers have argued for years about where to draw that line; the Zaporizhzhia case suggests the line may already have been crossed on a real battlefield.
The presence of a widely sold commercial AI module also raises hard questions about how off-the-shelf chips reach a sanctioned military, and whether any supplier can meaningfully control what its hardware ends up doing.
It matters because the debate over autonomous weapons has shifted from a hypothetical policy question to a forensic investigation of three deaths at a gas station.