Why Global Powers Removed Human Control From International Rules on Automated Weapons
Geneva, Sunday, 27 September 2026.
In closed-door UN negotiations, the United States and Russia stripped mandatory human oversight from a draft global pact, allowing future artificial intelligence weapons to strike targets without human review.
Diplomatic Shift in Geneva
In a decisive move during early September 2026, the United States and Russia successfully removed mandatory human oversight requirements from a draft global pact on lethal autonomous weapons [1][2]. This intervention occurred during United Nations-led negotiations in Switzerland, marking a significant pivot in the geopolitical debate surrounding military artificial intelligence applications [1]. The development signals that top global military powers are prioritizing tactical flexibility and autonomous defense development over international binding constraints [2]. Defense contractors and tech firms developing dual-use AI technologies are now facing a landscape with fewer international restrictions on autonomous defense systems [1].
The Negotiation Process
Hundreds of diplomats convened in a U.N. conference room in Geneva early this month to work toward regulating lethal autonomous weapons [2]. On the final day of negotiations, the talks turned aggressive as the two nations weakened a range of safeguards in the landmark effort [1]. The critical edits occurred during a single approximately 15-hour session where U.N. cameras were deactivated and civil society observers were excluded [2][3]. During this period, the U.S. and Russian delegations each deployed approximately 10 lawyers to the negotiations, significantly outnumbering other delegations and accelerating the editing process [2].
Erosion of Safety Protocols
The removed provisions included requirements for AI weapons systems to operate in a predictable and reliable manner, mandates for ethical considerations, and requirements for human review of AI-generated military targets prior to strikes [2]. The U.S. and Russia have collaborated to weaken international regulations on lethal AI weaponry, a process described by insiders as erosion piece by piece [2]. One person familiar with the talks stated it amounted to death by a thousand paper cuts as the two countries chipped away at protections [2]. The Washington Post reported that the U.S. and Russia specifically removed a provision that would have required humans to review AI-generated targets before a weapon carried out a strike [3].
Global Reaction and Framework
On September 26, 2026, it was reported that 128 UN member states reached an agreement on a non-binding framework regarding lethal autonomous weapons during negotiations in Geneva [4]. The final text of the framework was significantly diluted during the closing stages of negotiations due to opposition from the United States and Russia against stronger international restrictions [4]. While the agreement covers human judgment, control, predictability, reliability, traceability, and explainability, U.S. negotiators specifically advocated for greater flexibility and changes to language regarding predictability [4]. Both Washington and Moscow have consistently favored the implementation of national guidelines for autonomous weapons rather than the creation of a legally binding international treaty [4].
Support for Binding Treaty
Despite the dilution, a record 76 nations already back turning the nonbinding text into a binding treaty, according to Human Rights Watch [3]. This represents a specific portion of the agreeing states that desire stricter regulation. The proportion of states supporting a binding treaty relative to those who agreed on the framework can be expressed as 59.375 [3][4]. The non-binding framework is intended to serve as a basis for potential future formal negotiations on international rules for weapons capable of identifying and engaging targets with varying degrees of autonomy [4]. The UN Secretary-General and the International Committee of the Red Cross have issued warnings regarding the reduction of human control over lethal force and have advocated for stronger international restrictions [4].
Humanitarian Concerns
The UN’s humanitarian chief Tom Fletcher expressed deep concern, stating that AI weapons keep him awake at night [5]. He worries that the rules and values driving AI will make it even less humanitarian than humans are [5]. Human rights groups warned that the changes could let AI weapons select their own targets and reduced accountability for civilians who were harmed [3]. The U.S. is currently revising the Pentagon’s internal directive on lethal autonomous weapons, which was originally established in 2023 [2]. This approach to weakening international AI treaties may signal its forthcoming strategy for domestic regulation of AI military technology [2].