Artificial intelligence is reshaping international peace and security. It is generating new risks, from algorithmic targeting and disinformation to cyber operations and sanctions evasion, while also creating new opportunities for prevention, mediation, early warning, and peacekeeping.
The United Nations (UN) Security Council is increasingly aware of both dynamics. It has debated AI repeatedly, issued statements, and heard warnings from the UN Secretary-General and from leading figures in the private sector and academia. Most recently, on September 2, 2026, the Security Council held an informal interactive dialogue focused on practical applications of AI across conflict prevention and resolution. Yet it remains institutionally unwired to address either the risks of AI or its opportunities.
In military operations, AI expands capabilities: decisions to strike become automated or opaque while the risk of abdicating responsibility increases, complicating attribution and civilian protection. In civilian and hybrid domains, AI lowers barriers to cyberattacks, information manipulation, and other forms of disruption that often fall outside the Council’s traditional state-centric framework. At the same time, AI offers new tools for conflict prevention, mediation, peacebuilding, and peacekeeping. Across all three domains, the same governance problem emerges: the UN Security Council, the primary organ for the maintenance of international peace and security, remains poorly connected to the technological systems increasingly shaping conflict dynamics.
The problem is not simply one of technological change. It reflects a deeper mismatch between the Security Council’s current modus operandi and the nature of AI-enabled conflict. The Council was designed to respond to identifiable actors, attributable conduct, and threats to peace recognizable by its membership. AI-enabled conflict increasingly involves private companies, algorithmic systems whose functioning is difficult to observe, and digital infrastructure that blurs the boundaries of attribution. The Council’s existing subsidiary bodies and agenda items were not designed with AI-enabled conflict in mind, limiting the ability to respond effectively. This is particularly true when responsibility is disputed and civilian casualties are contested: technology facilitates faceless attacks, obscuring who is responsible and often confusing combatants and noncombatants.
This mismatch is reinforced by a growing institutional divide. Across the wider UN system, member states are constructing scientific panels, reporting mechanisms, and creating governance processes devoted to AI. The General Assembly has established recurring processes on AI in the military domain. New scientific and multistakeholder bodies are beginning to generate technical expertise and policy guidance. However, the Security Council has created no systematic way to draw on these developments, despite its primary responsibility for the maintenance of international peace and security.
In this paper, the authors highlight five practical gaps to explain this disconnection.
- The Council lacks structured access to technical expertise on AI-related risks.
- AI considerations are not consistently integrated into country-specific files.
- Existing mechanisms struggle to attribute responsibility for AI-enabled risks.
- The implications of AI for determining threats to peace under UN Charter Article 39, the trigger of Council’s action under Chapter VII, remain underdeveloped.
- The Council has not kept pace with the work of the wider UN system in using AI to strengthen its own work, including in prevention, mediation, peace operations, and sanction monitoring.
The Council was not designed with AI-enabled conflict in mind, and creating a dedicated subsidiary body or new agenda item would require a degree of consensus among the permanent members that does not currently exist. The most realistic path forward is, therefore, adaptation of the current architecture. Existing tools could be repurposed and used more effectively through recurring briefings on AI; consistent integration of AI-related risks into country-specific analysis; stronger use of expert mechanisms to monitor technology-enabled violations; clarification of how AI-enabled dynamics relate to threat determinations under Article 39 of the UN Charter and the subsequent actions of the Council under Chapter VII; and greater use of AI tools in prevention and peace operations, within appropriate safeguards.
The Council does not, for the most part, lack the machinery to engage AI but the political will to retool it. But these adaptations are not equally easy to pursue: regulating military AI will face resistance from the permanent members developing those capabilities, while using AI to strengthen the Council’s existing work in prevention, peacebuilding, and peace operations has emerged as an increasingly prominent and potentially less contested place to begin.
The Council’s significance, however, should not be measured only by its operational capacity. Its debates are already helping shape the normative environment surrounding AI and conflict. By elevating AI as a matter of international peace and security, providing a forum for states to articulate positions, and amplifying concepts such as “meaningful human control,” the Council is contributing to the gradual formation of customary international law and norms that may shape future governance frameworks. The repetition of these concepts across Council debates, General Assembly resolutions, and Secretary-General reports builds state practice and opinio juris (the two constitutive elements of customary international law). This normative accumulation makes certain positions harder to defend politically and creates conditions for eventual crystallization into treaty language.
