according to an exclusive report by new scientist, ukraine’s military application of artificial intelligence has reached a historic turning point: in 2024, during a covert operation on the bakhmut–chasiv yar front, ten quadrotor drones with full autonomous combat capabilities independently completed the entire process—target detection, identification, decision-making, and lethal strike—without any real-time human control, data transmission, or remote intervention. the drones confirmed the killing of multiple russian soldiers and the destruction of a military truck. this marks the world’s first technologically verified, traceable, and evidence‑based real‑world case of ai‑driven lethal autonomous weapons.
alexander kokhanovsky, the project lead and a ukrainian drone systems architect, admitted during a closed‑door briefing at the kyiv embassy: “launch means termination—it severs all communication links, transmits no imagery, requests no commands, and destroys whatever it sees.” these drones were preprogrammed for a flight range of 3–5 kilometers; upon reaching their designated area, they automatically activated local ai‑powered visual recognition and tactical decision‑making modules, operating entirely outside the rear‑echelon command network and achieving a truly “human‑out‑of‑the‑loop” lethality cycle.
post‑battle assessments conducted by manually piloted drones confirmed casualties and damage to equipment in the target zone. kokhanovsky clarified that this test was carried out independently by an undisclosed ukrainian frontline unit, and that he did not participate in the on‑site deployment. his aero center company was established after the experiment and maintains no organizational or legal ties to the operation.
compared with earlier sporadic reports—such as ukraine’s ai‑driven drones autonomously attacking russian armored vehicles in 2023 (without causing fatalities) or the 2021 un report mentioning turkey’s kargu‑2 allegedly conducting autonomous attacks in libya (lacking conclusive evidence)—this incident provides, for the first time, a complete operational log, detailed flight trajectory records, and third‑party battlefield verification, constituting the most robust technical and factual evidence to date.
although international law currently lacks a binding convention prohibiting lethal autonomous weapon systems, un secretary‑general antónio guterres has repeatedly emphasized that “machines should not have the power of life and death,” while oxford university digital ethics expert maria rosaria tadeo points out that such systems not only undermine humanitarian constraints on warfare but also render accountability mechanisms entirely elusive. notably, ukraine’s existing defense regulations still explicitly forbid ai from fully replacing human judgment in the final strike phase; however, the ministry of defense is urgently consulting with several defense‑technology firms to revise these provisions. meanwhile, professor anthony king of the university of exeter’s centre for strategic studies questions whether retaining human involvement in decision‑making is not merely a technological regression but rather a means of significantly improving target‑identification accuracy and tactical adaptability.
