U.N. Talks on Autonomous Weapons End Without Binding Rules Until at Least 2028
GENEVA — A week of United Nations-backed negotiations on weapons that can select and attack targets with little or no human oversight ended Saturday with an agreement to keep talking through 2028, but without a commitment to legally binding rules.
The outcome leaves the world’s major military powers free to continue developing increasingly autonomous systems while diplomats try to define the limits of acceptable use. The weapons are often called “killer robots,” though the formal debate covers a wide range of lethal autonomous weapons systems, including AI-assisted drones and targeting platforms.
Broad support, but no binding agreement
Representatives of the 128 states that are parties to the Convention on Certain Conventional Weapons met in Geneva beginning Monday. About 100 countries supported stronger, legally binding controls, according to Reuters. But the convention works by consensus, giving opponents the ability to block stronger language.
Major arms-producing countries including the United States, Russia and Israel resisted a binding mandate. The final decision calls on governments to “elaborate” a possible instrument over the next two years, leaving unresolved whether any future rules would carry legal force.
The Associated Press reported that seven countries — Austria, Belgium, Brazil, Ireland, Mexico, New Zealand and Panama — had proposed launching negotiations on rules that could prevent or limit machines from killing without meaningful human control or accountability. That push failed to win consensus.
Why the delay matters
Autonomous weapons are not a single, clearly defined device. The term can include systems that use sensors, software and artificial intelligence to identify, track and engage targets after activation. Supporters of regulation say decisions over life and death must remain under human judgment, with clear accountability when mistakes or abuses occur.
Concern has intensified as drones and other systems with autonomous functions appear in conflicts in Ukraine, Sudan and the Middle East. Advocates warn that battlefield technology is advancing faster than international law, potentially making meaningful safeguards harder to establish once the systems are widely deployed.
U.N. Secretary-General António Guterres, Pope Leo XIV and the International Committee of the Red Cross have called for strong, legally binding rules. Campaigners said the new mandate falls short of that goal, while U.N. disarmament chief Izumi Nakamitsu described the agreement as a significant step despite its limitations.
What happens next
Diplomats now have until 2028 to develop the outline of a possible international instrument. The hardest questions remain open: which systems should be prohibited, how much human control is enough, who is responsible for an autonomous strike, and whether the resulting framework will be legally binding or voluntary.
For now, the Geneva decision preserves a negotiating path but not a rulebook — a compromise that keeps nearly 100 countries’ push for stronger limits alive while allowing military AI development to continue without new global restrictions.