Sources: US and Russian diplomats worked to weaken an AI weapons pact at the UN this month, removing a requirement that humans review AI-generated targets, more

AI-controlled weapons can now be deployed without a human needing to approve each strike.

During UN negotiations this month, U.S. and Russian diplomats reportedly worked to weaken a potential pact regulating autonomous weapons. Closed-door discussions in Geneva saw the removal of provisions requiring AI weapons to operate predictably and reliably, language mandating ethical review, and a critical requirement for human review of AI-selected military targets before strikes. Observers described the effort as "death by a thousand paper cuts," with both delegations allegedly deploying larger legal teams to expedite these changes. Human Rights Watch expressed concern that this weakened language could permit machines to make life-or-death decisions without human oversight. This action occurred despite the increasing real-world use of AI targeting in conflicts like Gaza and the Iran war, and while the Pentagon is also revising its own directives on lethal autonomous weapons.

The U.S. and Russia's coordinated effort to dilute UN regulations on AI weapons signals a divergence in global approaches to military AI governance. While other nations and international bodies push for stricter controls and human oversight, these two major powers appear to be prioritizing operational flexibility over ethical safeguards. This move could accelerate the development and deployment of autonomous weapons systems, potentially leading to an arms race characterized by machines making critical decisions with less human intervention. The weakening of these regulations will likely influence how other countries draft their own domestic policies and international stances on lethal autonomous weapons.

This development directly impacts the landscape of international arms control and the future of warfare. By removing essential human review requirements, the U.S. and Russia are enabling a scenario where AI could independently select and engage targets, raising significant ethical and accountability questions. The absence of binding treaty language on predictability and reliability leaves open the possibility of unpredictable AI behavior in combat. This outcome may embolden states to pursue autonomous weapon capabilities with fewer international constraints, increasing the risk of unintended escalation and reducing the capacity for de-escalation in conflict zones.