UN Talks on Autonomous Weapons Collapse Amid Consensus Deadlock
A week of negotiations in Geneva ended without a legally binding mandate for autonomous weapons due to Russian objections and a lack of political will among major powers.
By Nia Foster ·

Efforts to establish a negotiating mandate for a legally binding international treaty on autonomous weapons have collapsed after week-long talks in Geneva failed to reach a final agreement. The conclusion came just before 5 a.m., following all-night sessions that could not overcome objections raised by Russia. While a significant majority of states at the UN Convention on Certain Conventional Weapons wanted to proceed with a binding framework, the requirement for unanimous consensus forced the conference to settle for a much weaker option.
The final mandate requested countries to elaborate an instrument by consensus, a process that may result in a document without strict legal obligations. Although Russia led the opposition using obstructive tactics, other military powers with significant investments in artificial intelligence warfare, including the United States, were also hesitant to commit to a binding protocol designed to protect civilians from robot drones and other autonomous systems.
Debate centered on whether existing international humanitarian law was adequate for the new challenges posed by these weapons. More than 100 states argued for new legislation, while European and NATO states, along with China and Iran, expressed support or a willingness to pursue a new protocol. The United States noted that the need for human control over autonomous weapons had been discussed during this week’s APEC summit meetings between President Donald Trump and President Xi Jinping, yet it remained reluctant to agree to a binding treaty.
Several defense technology powers, including South Korea, Japan, Turkey, the UK, France, and Germany, resisted being bound by new law. These nations advocated for a nonbinding agreement instead. The UN forum operates on a consensus model, meaning all 128 parties must agree unanimously to pass a statement. This high threshold has prevented the convention from passing any new weapons regulations in the past 30 years, leading states to seek alternative pathways for treaties, such as the 2008 ban on cluster munitions.
Max Tegmark, chair of the Future of Life Institute, stated that world leaders had failed to meet the moment. He remarked that slaughterbots are already present and that states had agreed to merely discuss the issue for the 14th year. Diplomats inside the negotiations reported that Russia had deliberately filibustered and not acted in good faith, contrasting with US delegates who were seeking a path toward a consensus statement.
The final day of the conference saw a rare protest in the chamber, triggered by the exclusion of several nonmember countries from negotiations. The Egyptian delegation shouted at the conference chair during this disruption. Critics highlighted that the outcome represented a missed opportunity, noting that the issue had been discussed for 13 years and that large parts of a potential treaty had already been settled.
A European diplomat commented that there was no political will to truly regulate these weapons in a manner commensurate with the urgency of the need. Bonnie Docherty, a senior arms adviser at Human Rights Watch, emphasized the necessity for binding law. She noted that treaties could be negotiated within 15 months and dismissed the US proposal of five years to sort out a mandate.
A bloc of seven countries, including Austria, Belgium, Brazil, Ireland, Mexico, New Zealand, and Panama, had led the push for a binding treaty to be negotiated within one year. Brazil’s delegate told the floor on Wednesday that these were not weapons of the future but systems already in use and further developed. The delegate concluded that the international community was late in developing the necessary norms and rules to regulate lethal autonomous weapons systems.
The backdrop for these negotiations included the use of AI-enabled targeting systems by Israel and the United States in operations in Gaza and Iran, which resulted in mass casualties. Reports indicate the Pentagon struck more than 1,000 targets in Iran on the first day of the war. Investigative media reports also suggest the US military aims to carry out 1,000 AI-assisted targeting decisions per hour, underscoring the practical reality of the technology being debated in Geneva.






