As artificial intelligence spreads into defence systems, the United Nations is trying to give the world a place to discuss the risks. Its disarmament research arm is launching what it describes as the first international forum on military AI, As artificial intelligence spreads into defence systems, the United Nations is trying to give the world a place to discuss the risks. Its disarmament research arm is launching what it describes as the first international forum on military AI, according to the Associated Press.
What is the UN’s new military AI initiative?
The UN Institute for Disarmament Research (UNIDIR), based in Geneva, is setting up a forum to bring together actors from every relevant sector worldwide to work on AI in military uses. It sits inside UNIDIR’s Center of Excellence on AI, Peace and Security, created this summer as the UN’s first permanent platform on these issues. UNIDIR director Robin Geiss said the center received funding from all five major nuclear powers: the US, Russia, China, Britain and France.
Geiss was vague about exactly what the forum will do. He said it is one of several AI efforts the center is launching this week.
Why UNIDIR is worried: shrinking room for human judgment
Geiss warned that the room for human judgment is getting narrower, including in the nuclear domain. He said it is hard to judge how close the world is to machines acting outside human control, and that this uncertainty is the core problem the initiative wants to address.
The difficulty is that AI is already entering many areas of military activity, including early warning systems, information gathering and reconnaissance. That makes it hard to say when meaningful human control is lost.
Three tracks to watch
1. The international forum
A multi-stakeholder space for governments, civil society, academia, experts and industry to discuss AI in military use.
2. A framework for responsible industry behaviour
Geiss said design and development decisions by companies affect whether AI-enabled military systems are reliable, predictable and consistent with international law. UNIDIR is developing practical, voluntary guidelines grounded in international law and human rights. Consultations this week will involve defence technology firms, software developers, subcontractors and investors.
3. Risks from armed groups
UNIDIR is also launching work on misuse of AI by armed groups. Geiss said such groups are already using the technology to support operational planning, execution and logistics.
Geopolitics: rivals, but shared concern
Geiss acknowledged that the global race to lead in AI creates serious obstacles. But he said concern about AI’s impact now cuts across geopolitical blocs and is prompting action from major powers. As an example, he pointed to a proposal last month for the US and China to create an AI safety alert system for issues that could threaten national security.
AI risks were also a major theme in speeches by leaders of many of the UN’s 193 member states at last month’s annual high-level meeting.
What this means
- Voluntary, not binding: The industry framework is a set of voluntary guidelines, not a treaty or enforceable rule.
- Early stage: Details of the forum’s work remain unclear.
- Companies are now part of the conversation: The focus on developers and investors reflects that choices made in labs and boardrooms shape how military AI behaves.
- Nuclear risk is central: The concern is not only futuristic autonomous weapons, but also AI quietly influencing early warning and decision-making.
Frequently asked questions
What is UNIDIR?
The UN Institute for Disarmament Research, a Geneva-based UN institute that researches disarmament and security issues.
Is the UN banning military AI?
No. This is a forum and a set of voluntary guidelines, not a ban or binding treaty.
Who funds the Center of Excellence on AI, Peace and Security?
According to UNIDIR’s director, funding came from the US, Russia, China, Britain and France.
Why does AI in early warning systems matter?
Because errors or over-reliance could reduce meaningful human judgment in high-stakes decisions, including nuclear ones.
Source: Associated Press, UN initiative on military AI