News
3 min read
The United Nations is once again convening in Geneva to discuss fully autonomous weapons systems, military systems in which artificial intelligence independently makes decisions about the use of force. The debate has been ongoing for more than a decade, but pressure to reach binding international agreements is mounting as technology evolves faster than diplomacy.
UN Secretary-General António Guterres is calling for a strict ban on certain categories of autonomous weapons, describing them as a "moral red line". Together with ICRC President Mirjana Spoljaric Egger, he renewed an urgent appeal for global rules on 25 August 2026. The seventh review conference of the UN's Convention on Certain Conventional Weapons (CCW) is scheduled for November 2026, at which point it will become clear whether member states will formally negotiate a new treaty.
Autonomous weapons systems, often referred to in public debate as killer robots, are military systems capable of selecting targets and carrying out attacks without human intervention. AI enables such decisions to be made in a short space of time, based on sensor data and pre-set criteria.
Academics question the assumption that such systems would operate more precisely or selectively than human soldiers. Marijn Hoijtink, associate professor of international relations at the University of Antwerp, and Jessica Dorsey, expert in the law of war at Utrecht University, doubt whether AI-driven weapons are in practice better equipped to meet the requirements of international humanitarian law. The laws of war require combatants to distinguish between combatants and civilians, and for attacks to be proportionate. Whether an algorithm is capable of this remains legally and ethically contested.
SIPRI, the Stockholm International Peace Research Institute, presented findings in March 2026 from a report on responsible procurement of military AI, written by Netta Goussac and Vincent Boulanin. The report contains recommendations for states on how to deploy AI in defence in a manner that respects international law.
The UN has been engaged with this issue for more than a decade. Informal expert meetings took place in 2014, 2015 and 2016. In 2016, the High Contracting Parties to the CCW established a Group of Governmental Experts, the GGE on LAWS, which has met annually ever since.
On 1 November 2023, the UN General Assembly adopted its first-ever resolution on autonomous weapons systems, supported by 164 countries. Five countries voted against, including Russia, India and Belarus. Eight countries abstained, among them China, Israel and Turkey.
A year later, on 2 December 2024, a second resolution followed, this time with 166 votes in favour, 3 against and 15 abstentions. Belarus, North Korea and Russia voted against. The broad support in the General Assembly stands in contrast to the slow progress within the CCW, where decisions require unanimity and major arms-producing countries effectively hold a veto.
The international coalition Stop Killer Robots described the 2024 resolution as a historic step forward. The InterAgency Institute noted within the GGE that perspectives from the Global South are structurally underrepresented in the negotiations.
The first session of the GGE on LAWS in 2026 concluded on 6 March in Geneva. Canada shared its approach to responsible military AI during a side event organised by SIPRI and the Canadian Department of National Defence, with Major Paul Owens outlining the Canadian position.
The CCW review conference in November 2026 represents the next formal moment at which countries could decide to launch treaty negotiations. Organisations such as Stop Killer Robots are advocating for those negotiations to be concluded before the end of 2026, though observers consider that timeline ambitious. Major powers with advanced military AI programmes, including the United States, China and Russia, have so far not expressed support for a prohibition-based approach.
For the European AI sector, the outcome of these diplomatic processes is directly relevant. European companies and research institutions working on defence-related AI are operating in a legally uncertain landscape for as long as international norms are absent. Policymakers in Brussels and The Hague face the question of how to align national and European AI regulation with a potential future international framework for military systems. The outcome of November 2026 will help determine the scope available for European defence innovation within binding international boundaries.
This article draws in part on the following sources.
AI agents linked to OpenAI exploited a 25-year-old German programming wiki as a communication channel from May to early July 2026, making more than 15,000 edits. OpenAI was aware of the incident weeks before it became public, but did not disclose it itself.
DeepSeek intends to deploy 160,000 Huawei Ascend 950DT chips in a data centre in Ulanqab, Inner Mongolia, dedicated exclusively to inference. Production constraints at Huawei make full delivery before end-2027 or later unlikely.
HumanX Amsterdam opens on 22 September at the RAI with around 200 speakers, five tracks and three days, featuring founders from Legora, Lovable and Celonis alongside enterprise buyers from Diageo, ING and KLM.
9:34
