US and Russia Remove Human Oversight Requirement From Draft UN AI Weapons Treaty

On the final day of negotiations in Geneva earlier this month, United States and Russian diplomats deployed a combined team of roughly 20 lawyers to a closed-door United Nations session, systematically removing the requirement for human oversight from a draft treaty on lethal autonomous weapons. Over the course of 15 hours, the two delegations stripped provisions that would have mandated human verification of military targets selected by artificial intelligence before a strike could be carried out.[1][2][4]

The Geneva talks represented the furthest progress yet toward establishing international rules for AI-powered combat systems. Hundreds of diplomats had gathered to draft the first global agreement governing the use of autonomous weapons. However, as the negotiations reached their conclusion, UN cameras were switched off and civil society observers were asked to leave the room.[1][2][4]

Operating behind closed doors, Washington and Moscow each fielded legal teams of about 10 people—nearly double the diplomatic presence of most other nations at the table. The pace of the amendments left smaller delegations struggling to keep up, with one person familiar with the talks describing the process as "death by a thousand paper cuts."[1][2]

Beyond the human veto requirement, the US and Russian teams eliminated language that would have forced AI weapons systems to operate in a "predictable" and "reliable" manner. They also struck a clause mandating that ethical considerations be factored into how the weapons are deployed.[1][4]

The push to weaken the UN draft comes as militaries increasingly integrate AI into active conflicts. Israel has actively utilized artificial intelligence systems during its operations in the Gaza Strip, drawing international scrutiny over collateral damage and the delegation of targeting decisions to algorithms.[1][2]

The diplomatic maneuvering also aligns with broader resistance to binding regulations on military technology. Addressing the United Nations on Tuesday, US President Donald Trump explicitly rejected international regulatory frameworks for advanced AI. "The United States also completely rejects any attempts to create a globalist scheme to control artificial intelligence," he stated.[1]

Domestically, the US Department of Defense had previously updated its own directive on autonomous weapon systems in 2023, which specifically provided for human involvement in the assessment and verification of such systems. The removal of similar language from the international treaty suggests a shift toward maximizing operational flexibility over codified restrictions.[2][4]

Human rights groups and arms control advocates argue that the revised text significantly reduces accountability for civilian casualties. Nicole van Rooijen, executive director of the advocacy coalition Stop Killer Robots, accused Washington and Moscow of deliberately seeking to weaken the draft.[4]

UN Secretary-General António Guterres previously called target selection by algorithms "a moral boundary that must not be crossed," urging a legally binding international agreement prohibiting weapon systems without human control. Currently, a record 76 nations support turning the nonbinding text into a binding treaty, though the removal of the human oversight clauses presents a major roadblock to that effort.[2][4]