US and Russia strip human control and ethical rules from UN draft treaty on AI weapons
During closed-door UN negotiations in Geneva, American and Russian delegations worked across 15 hours to remove mandatory human verification and ethical safeguards from the draft agreement on autonomous weapons.
Closed-door revisions in Geneva
Hundreds of diplomats convened in Geneva in early September 2026 to negotiate the first international framework governing the use of lethal autonomous weapons. On the final day of the proceedings, negotiations grew aggressive during an unbroadcast 15-hour session dedicated to amending the draft text. Organizers turned off United Nations cameras and excluded civil society observers from the conference room. United States and Russian delegations each deployed approximately 10 lawyers, roughly double the representation of other participating states, to introduce extensive textual alterations. Delegations from smaller countries were unable to track the volume and speed of edits presented by the two powers.
Stripped safeguards and oversight
The bilateral modifications eliminated foundational requirements designed to constrain artificial intelligence in warfare. Negotiators removed language requiring autonomous weapons systems to operate in a predictable and reliable manner. The American and Russian teams also removed clauses mandating that militaries weigh ethical considerations when using AI systems. Furthermore, the revised text excluded a requirement for human operators to inspect and verify military targets identified by artificial intelligence before launching an attack. Verity Coyle, deputy director for arms at Human Rights Watch, stated that weakening these standards shifts life-and-death choices to machines, increasing civilian risks and lowering accountability for potential war crimes.
machines will be able to make life-and-death decisions without human control
United Nations High Commissioner for Human Rights Volker Türk likewise reiterated his backing for strict limitations on automated targeting systems during the September sessions.
ban weapons that can kill without human intervention
Operational integration and military policy
The revisions in Geneva parallel the expanding role of automated technology across modern military operations. The United States military deployed an artificial intelligence model developed by Anthropic to identify nearly 1,000 targets within the initial 24 hours of operations in Iran. United States officials are also in the process of revising the military's own 2023 lethal autonomous weapons directive. In the Middle East, Israeli armed forces have utilized artificial intelligence for surveillance and target selection in the Gaza Strip. In Eastern Europe, Russian and Ukrainian forces have both deployed lethal autonomous drones in ongoing combat.
- Pentagon releases its directive on lethal autonomous weapons
- UN General Assembly adopts resolution on autonomous weapons with three opposing votes
- US and Russia remove human verification and ethical rules during Geneva draft treaty talks
- Participating states scheduled to resume UN talks in Geneva
Treaty status and upcoming negotiations
The United Nations negotiations do not currently possess binding legal authority, as participating governments determine whether the draft can become an official mandate or treaty. The United States Department of State, the Russian Foreign Ministry, and the United Nations did not respond to requests for comment regarding the Geneva session. Stop Killer Robots coalition director Nicole van Rooijen pointed out that Moscow and Washington resist binding legal frameworks despite exerting the greatest influence over the talks. The current process follows a December 2024 UN General Assembly resolution on autonomous weapons, which passed with only Russia, Belarus, and North Korea voting against it. Delegations are scheduled to return to Geneva in November 2026 for the next round of negotiations.
