The Pulse
UNIDIR Opens a New Forum on Military AI at UN Headquarters
UNIDIR is launching a multistakeholder forum on artificial intelligence, peace and security at United Nations Headquarters in New York on October 22, 2026. The launch coincides with the opening of expressions of interest to join the forum,
AI.info Team · Written with AI · fact-checked ·
UNIDIR is opening a new multi-stakeholder forum on artificial intelligence in international peace and security at United Nations headquarters in New York on Thursday, October 22. The event, organized with Switzerland and Pakistan, also marks the formal opening of expressions of interest in the forum. UNIDIR says the platform will bring continuity to discussions that span military AI, peace and security.
The forum sits within UNIDIR’s Centre of Excellence on AI, Peace and Security, launched in Geneva in June. Its creation comes as governments face questions about AI’s expanding role in military operations and how to manage risks without relying on a single governance process. The forum is intended to connect governments, industry, civil society, researchers and other specialists.
Centre initiatives accompany the forum launch
UNIDIR Director Robin Geiss told reporters on October 20 that the forum will engage “actors from every relevant sector and all over the world.” Alongside the forum, Geiss described three other Centre initiatives. One concerns a framework for responsible industry behaviour, with practical, voluntary guidance for companies developing military AI. UNIDIR says the guidance will be grounded in international law and the UN Guiding Principles on Business and Human Rights; development and consultations are already underway.
A second initiative will develop an AI security lexicon. Geiss said governments and other stakeholders often use terms such as “autonomy,” “safety” and “security” differently, making it harder to establish shared understandings. The project is intended to clarify key concepts while recording where views still diverge.
The third initiative concerns the possible misuse of AI by non-state armed groups, including support for operational planning, execution and logistics. Geiss said the Centre is examining where existing non-proliferation and export-control approaches leave gaps, while seeking measures that do not restrict peaceful access to AI. The institute has not presented the forum as a treaty negotiation or a body with enforcement powers.
Human judgment is the central concern
In the October 20 briefing, Geiss warned that military decision cycles are compressing and that the margin for human judgment is narrowing, including in the nuclear domain. He pointed to AI applications in intelligence analysis, target identification, drones, cyber operations and information operations. He also cautioned that opaque systems can increase the risk of error, miscalculation and unintended escalation.
“Decision cycles are compressing.”
Robin Geiss, Director of UNIDIR, at a UN press conference on October 20, 2026
The UN transcript records Geiss describing uncertainty about when AI becomes embedded in military processes as part of the problem: it can be difficult to identify the point at which human control is being lost. He said the Centre’s work will include helping states develop national approaches to AI in security and defence.
Voluntary guidance, not a new treaty
The Centre’s industry framework is aimed at company decisions across system design, training data, testing and support during and after deployment. UNIDIR argues those choices affect whether AI-enabled military capabilities are reliable and predictable and whether they can operate in line with international law. Its planned guidance is voluntary; the announcement does not establish binding rules for states or companies.
UNIDIR’s event calendar lists further public discussions for Friday, October 23, on the industry framework and a security lexicon for AI. Those sessions are the next scheduled steps in a programme designed to gather views from governments, companies, civil society and researchers. The October 22 event launches the forum as a continuing platform for engagement; the Centre’s separate initiatives were already underway or being introduced during the same week.