News and context with references. Check the publication date and source dates.
Artificial intelligence has been widely used in military operations, while the role of human judgment is rapidly diminishing. According to an UN News report on October 20, 2026, Robin Geiss, Director of the United Nations Institute for Disarmament Research, or UNIDIR, warned that this development could allow machines to increasingly dominate decisions on the battlefield.
AI's role continues to expand
During a briefing at UN Headquarters in New York on Tuesday, Geiss questioned whether machines would be allowed to select targets and kill people. He said AI is increasingly on a path toward displacing human decision-making in military affairs.
The use of AI in this sector includes intelligence analysis, target identification, drone operations, and the coordination of cyberattacks. According to Geiss, the growing responsibilities assigned to these systems are coming at the expense of human judgment, including in the nuclear domain.
The UN Secretary-General has also repeatedly urged stakeholders not to cross this moral red line. He has called for stronger regulation of military AI and a legally binding treaty on lethal autonomous weapons systems. Geiss said the international community is increasingly responding to that call.
Operational speed and the risk of escalation
The shift toward the use of AI is increasing the speed of battlefield operations. The time between identifying a target and carrying out an attack is becoming shorter, while forces can engage an unprecedented number of targets in a single day.
However, many evolving AI systems remain opaque. The head of UNIDIR warned that this situation increases the likelihood of errors and miscalculations, which could ultimately trigger unintended escalation.
The accessibility of AI tools also poses another challenge. Geiss said it will become increasingly easy for non-state actors to conduct more sophisticated military operations. The risk of misuse could include support for the planning, execution, and logistics of operations by non-state armed groups.
Geopolitical obstacles and opportunities for cooperation
Geiss acknowledged that geopolitical competition and tensions pose serious obstacles to regulating the involvement of AI in the military sector. Nevertheless, concerns about its risks are increasingly shared by parties with differing political views, including the world's major powers.
For example, in September, the United States proposed an AI notification mechanism to China. The mechanism was intended to allow the two countries with the world's largest military expenditures to warn each other in the event of an AI incident that could threaten national security.
This development indicates that there is room to establish shared safeguards, even though countries' strategic interests remain different. In this context, clarity of terminology, communication about incidents, and a common understanding of the limits on AI use are considered important for supporting trust and cooperation.
A new forum within the UN system
UNIDIR is preparing several foundational elements for multilateral cooperation. On Thursday, the UNIDIR Centre of Excellence for AI, Peace and Security is scheduled to launch a multistakeholder forum on AI in international peace and security and the military domain. According to Geiss, the forum is the first and only one of its kind within the UN system so far.
One of its focuses is to develop a framework for responsible industry conduct, consisting of practical and voluntary guidance for companies during the design and development of military AI tools. UNIDIR is also developing an AI security lexicon to clarify concepts such as autonomy, safety, and security, which are currently used differently by various parties.
Differences in the use of terminology are considered to create ambiguity that hinders trust and cooperation. In addition, the forum will discuss the risk of AI misuse by non-state armed groups to support various aspects of their operations.
