By Sean Flynn

A Russian drone approached a gas station in Zaporizhzhia, Ukraine, on July 6, 2026. As it attempted to reach its target, the aircraft crashed into a wall and exploded, sending shrapnel into the surrounding area. Three civilians were killed, including 19-year old student Tetiana Bubynets, who had attempted to flee the approaching drone.

According to an August investigation by The New York Times, Ukrainian military officials and investigators determined that the drone had been equipped with an experimental artificial intelligence system capable of selecting its final target without human intervention. Although Russian operators programmed the drone to approach the gas station, its onboard software reportedly identified propane tanks as the intended target. Reflecting on this development, Ukrainian air defense commander Col. Serhiy Minaiev told The New York Times, “Machines are making decisions to strike.”

The incident reflects the increasing ability of autonomous weapons to identify and attack targets independently, an emerging development in modern warfare that raises questions about military effectiveness and civilian safety.

Drones have become central to the war in Ukraine, where both sides use them for surveillance and attacks. Unlike traditional remotely piloted drones, which are vulnerable to disrupted communications, autonomous drones offer potential military advantages through independent targeting and the ability to operate without continuous human control.

According to the International Committee of the Red Cross (ICRC), autonomous weapons can independently select and attack targets after being activated. Additionally, not all autonomous weapons use AI, and not every AI-equipped weapon is autonomous.

In Zaporizhzhia, Ukrainian investigators recovered drones containing Nvidia Jetson Orin minicomputers. These minicomputers can support AI image recognition, and investigators reported finding software trained to recognize specific objects, including propane tanks. These findings, along with the fact that there was an absence of radio antennas, support their conclusion that the drones could navigate and identify targets independently, without continuous communication with human pilots.

One major advantage of autonomous targeting is its potential resistance to electronic warfare. Remote-controlled drones rely on communication links that enemy forces can disrupt through electronic interference. Autonomous systems can instead use onboard computers to navigate and identify targets, potentially allowing them to operate when those links are unavailable.

Supporters of this technology also argue that AI-based targeting could improve precision and actually reduce civilian harm. Autonomous weapons are equipped with cameras and sensors that can recognize certain objects and targets before striking. Ukraine has experimented with similar technology, reportedly testing self-targeting drones against military equipment and fuel storage facilities in Russian-occupied Crimea. According to Ukrainian officials cited by The New York Times, no civilians were killed during those particular tests.

However, increasing autonomy introduces new risks, as evident from the Zaporizhzhia attack, which demonstrated how autonomous targeting systems may still contribute to civilian casualties even when their software is reportedly attempting to strike an object rather than a person.

The ICRC warns that autonomous weapons operating across larger areas and against broader categories of targets may become increasingly difficult to predict and control. As the technology currently stands, it can recognize objects and identify a target. However, the technology lacks the ability to properly determine whether attacking a target complies with international humanitarian law. The ICRC emphasizes that such context-dependent legal judgments “cannot simply be ‘coded in’ in advance.” Militaries must distinguish civilians from military objectives and consider the potential consequences of attacks on surrounding populations.

These concerns have intensified calls for international regulation in this technology. In August 2026, the United Nations Secretary-General Antonio Guterres and ICRC President Mirjana Spolijaric renewed their appeal for legally binding rules governing autonomous weapons. The ICRC has called for prohibitions on unpredictable autonomous systems and those designed to target people directly, alongside restrictions on other autonomous weapons to address humanitarian and legal concerns.

Governments will have an opportunity to address these issues at the Seventh Review Conference of the Convention on Certain Conventional Weapons, scheduled for November 16-20, 2026.

The Zaporizhzhia attack illustrates how autonomous targeting has begun moving from experimental development into active warfare. As Russia, Ukraine, and other militaries continue exploring the technology, its potential advantages will have to be weighed against unresolved questions of safety and human oversight. As autonomous weapons become more capable and accurate to operate, the dilemma of whether they should be allowed to make targeting decisions without human oversight becomes more apparent. 

Leave a Reply

Trending