Legal scholar urges binding treaty on autonomous drones

A legal scholar writing in Tech Policy Press argues that international humanitarian law can no longer absorb what autonomous drones are already doing on the Ukrainian front, and wants states gathering at the UN in Geneva next week to negotiate a binding treaty.

León Castellanos-Jankiewicz, a senior researcher in international law at the University of Amsterdam Faculty of Law and non-resident fellow at the UN Institute for Disarmament Research, opens on two incidents. In one, he writes, "A swarm of ten machines slammed into a field near Bakhmut, ending the lives of at least two combatants." This past July, a Russian drone guided by experimental AI killed three Ukrainians near a gas station in Zaporizhzhia. He quotes Ukrainian drone maker Alexander Kokhanovskyy on how the targeting works in practice: "We just launch it and we know everything will be dead" in the selected area.

The gap he presses on is that existing law assumes a human soldier choosing a specific target. In the Zaporizhzhia strike, he writes, "The drone operators selected a target category, not a victim." Once drones are flying in jammed conditions, "it becomes impossible to verify what the drones saw or targeted."

Castellanos-Jankiewicz then points at the industry layer. Anthropic has said Russian developers used AI to build kamikaze attack drone software, and a commercially available Nvidia chip reportedly steered the drone in Zaporizhzhia; defense companies have put $4.1 billion into tech start-ups. The International Committee of the Red Cross has called for autonomous weapons to be "prohibited outright." The draft report on the table at the UN Convention on Certain Conventional Weapons review conference, in his reading, "assigns no responsibility to the tech companies building these systems, who fiercely protect their commercial secrets." Two researchers we track circulated the piece on publication day.