US and Russia strip safeguards from UN AI weapons treaty

United States and Russian diplomats collaborated to remove critical safeguards from a draft United Nations treaty intended to regulate lethal autonomous weapons systems. During closed-door negotiations in Geneva, Switzerland, the two powers successfully stripped several key provisions from the proposed framework.

The removed language included requirements for AI weapons to operate in a “predictable” and “reliable” manner, as well as mandates for ethical considerations during deployment. Most significantly, the diplomats eliminated a provision that would have required human oversight to verify military targets identified by artificial intelligence before an attack could be launched.

Reports indicate that the changes occurred during a 15-hour period on the final day of talks. During this time, UN cameras were turned off and civil society observers were excluded from the room. Both the U.S. and Russian delegations utilized large legal teams—approximately ten lawyers each—to push through the amendments, which reportedly overwhelmed smaller delegations.