Ukrainian drone maker says autonomous drones killed soldiers in a test

In short: A Ukrainian drone manufacturer says it ran a one-time battlefield test where drones attacked targets without a human controlling the final decision.

What happened

Alexander Kokhanovskyy, CEO of Ukrainian drone maker Aero Center, told New Scientist that his company tested quadcopter drones that were programmed in advance to fly to a front-line area. Once there, the drones switched into what he called an AI-powered “Terminator mode.” In plain terms, this means the drones were meant to search an area on their own and attack targets without a person guiding them moment by moment.

Kokhanovskyy said there was no live video feed or other direct recording showing what the drones attacked. He said human-piloted drones later checked the area and found “a couple” of dead Russian soldiers. Based on that, he said the team concluded the autonomous drones had killed them.

A Ukrainian military commander told New Scientist that his unit uses only semi-autonomous tools, meaning a human still makes key choices like picking a target and starting an attack. He also pointed to Ukraine’s stated commitment to international humanitarian law, including taking care to avoid civilian casualties.

Why it matters

This claim matters because it highlights a hard line that many people worry about, weapons choosing and attacking human targets without a human’s final approval. Even if this was a one-time test, it raises questions about safety and responsibility, like the risk of “friendly fire” or hitting civilians. It also comes as Ukraine and Russia keep adding more automation to drones, often for navigation when signals like GPS are jammed (like trying to drive when your map and phone reception stop working).

Source: Arstechnica