Autonomous Weapons Systems Mini-Symposium: Beyond Yes or No – The Decisions States will Make this Year on Autonomous Weapons

[Netta Goussac is Associate Senior Researcher at the Stockholm International Peace Research Institute (SIPRI) and Special Counsel at Lexbridge.]

Editorial Note: This post forms the third of three posts of the Opinio Juris mini-symposium on the upcoming meeting of the Group of Governmental Experts on Lethal Autonomous Weapons Systems within the auspices of the Convention on Certain Conventional Weapons. A version of these reflections was delivered at the ASEAN Regional Forum Inter-Sessional Meeting on Non-Proliferation and Disarmament in April 2026.

A Decisive Year

This year will be an inflection point for the international regulation of autonomous weapon systems (AWS). The three-year mandate of the Group of Governmental Experts on lethal autonomous weapons systems (GGE), convened under the Convention on Certain Conventional Weapons (CCW) and tasked with formulating, by consensus, a set of elements of an instrument, will conclude at the end of the forthcoming meeting from 31 August to 4 September 2026. Its final report is due to the Seventh Review Conference of the CCW in November. States will then be invited to decide what, if anything, should come next. The choices they make will shape not only the governance of AWS but may also influence how the international community approaches the regulation of military technologies more broadly, including artificial intelligence.

That much is widely agreed. But the public framing of the decision ahead – should states negotiate a legally binding treaty on AWS, or not? – understates what is actually happening in the GGE meeting room.

Drawing on research published by SIPRI in September 2025 in the report Towards Multilateral Policy on Autonomous Weapon Systems, co-authored with Alexander Blanchard, this post sets out three observations about the current state of the debate. The key point is that the range of constructive options available to states in 2026 is wider than the public conversation implies, and that the opportunities for constructive engagement are greater than the pace of negotiations might suggest.

The Considerations Beyond the Binary

Through interviews with state representatives from all UN regional groups, and through analysis of official documents, the research identified a range of considerations that diplomats are weighing as they form their positions. These include: the scope of possible policy responses; the likely real-world impact of any instrument; timing; coherence with existing legal and governance frameworks; whether to pursue a single instrument or proceed incrementally; geopolitical and domestic political contexts; the legitimacy of both process and outcome; and the breadth of support that any outcome would command.

These considerations are not always articulated openly, for understandable reasons of diplomatic sensitivity and procedural constraint. When such factors remain unspoken, the debate can appear more polarised than it is, and opportunities for constructive engagement may go unnoticed. A state that has reservations about the timing of a binding instrument is not necessarily opposed to its substance; a state that advocates for a treaty may nonetheless recognise that interpretive guidance, transparency measures and practical operational guidance will be needed alongside it.

On this basis, it may be helpful for states to develop positions across the full range of policy directions, rather than only on the treaty question. In our report we identify tools that can operate alongside, or in place of, a legally binding instrument: political statements; interpretive guidance on existing international law; practical guidance for militaries and weapon developers; the integration of AWS-related questions into the agendas of other relevant forums; and transparency and confidence-building measures. The underlying observation is that no single tool is likely to address every concern, and that a coherent response will probably involve several.

More Common Ground than the Slow Pace of Discussions Suggests

The extent of common ground among states appears to be larger than the pace of CCW negotiations would suggest. There is broad support among participating states for the CCW as an appropriate forum. There is widespread endorsement of what has become known as a two-tier approach, distinguishing between autonomous weapons whose use would be unlawful in all circumstances, and those that should be subject to limits and regulation. And there is wide agreement that international humanitarian law provides the primary legal framework.

Recent developments support this reading. At the September 2025 session of the GGE, Brazil delivered a joint statement on behalf of 42 states affirming readiness to move toward negotiations on the basis of the Chair’s rolling text; this number increased to over 70 in during the March 2026 session of the GGE (more than half of the CCW High Contracting Parties). In December 2025, a UN General Assembly resolution on autonomous weapons that called on CCW parties to work towards elements of an instrument with a view to future negotiations was supported by 156 states. These numbers suggest that the slow pace of progress cannot easily be attributed to disagreement on substance.

The obstacles appear to lie elsewhere – in the broader geostrategic environment, in the security concerns of individual states, and in domestic political pressures. In other words, the deficit may be one of political will and leadership, rather than of technical consensus.

This carries a hopeful implication for the period ahead. The GGE has developed a rolling text: a working document that reflects contributions from a large and diverse group of states. Whatever direction states ultimately choose, this text is likely to serve as a reference point. The research identifies several avenues through which it may be possible to preserve and build on the text, including detailed statements at the GGE and the First Committee of the UN General Assembly (dealing with disarmament and international security), the sharing of national positions, and regional discussions aimed at building mutual understanding. 

The International Policy Debates on Military Artificial Intelligence 

AWS and the distinct and broader category of military artificial intelligence are often treated as distinct topics, handled in distinct forums. The AWS debate sits principally at the CCW; military AI questions have moved through venues including the First Committee, the Summit on Responsible AI in the Military Domain (REAIM), and the US-led Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy. The underlying concerns, however, overlap considerably. Both AWS and broader military AI touch on the role of humans in decisions to use force. Both raise questions of algorithmic bias, reliability and predictability. Both raise difficult questions of accountability and transparency.

States have, understandably, sought to avoid duplication between the two sets of discussions. In practice, however, those discussions have tended to proceed in parallel, with relatively little exchange between them. This carries some risk of fragmentation: governance frameworks for military AI and for AWS could evolve in ways that are inconsistent, or that leave gaps.

In the SIPRI report we identify opportunities for each process to learn from the other. The GGE, the First Committee and REAIM may each offer space for that kind of reflection, as may the Centre of Excellence on AI, Peace and Security launched by the UN Institute for Disarmament Research in June 2026.

What 2026 asks of States

Multilateral policy on AWS remains a test of the international community’s ability to govern emerging technologies of warfare collectively. Difficulties in conceptualising the concerns that AWS raise, inherent limitations in the processes and tools available for multilateral efforts, and eroded trust among states have stymied policy responses over more than a decade. The choices made this year will shape not only the governance of these weapons but also, likely, the broader trajectory of military AI governance.

Those choices are not simple. But they appear to be less binary than the public framing sometimes implies. States have a wider range of tools at their disposal, and more room for constructive engagement, than a yes/no debate would indicate. 

A useful question for delegations preparing for the upcoming session of the upcoming session of the GGE, the 2026 meeting of the First Committee and the CCW Review Conference may therefore be not only what position they take on a binding instrument, but what combination of approaches – binding and non-binding, substantive and procedural, at the CCW and beyond – they would support, and how those policy choices relate to the parallel conversation on military AI. Posed that way, the question 2026 asks of states is less about a single decision and more about the governance architecture they are prepared to build.