UN Urges AI Oversight in Warfare as Pakistan Flags Emerging Risks
WASHINGTON: Artificial intelligence is expanding its role in warfare at the expense of human judgement, the head of the United Nations’ independent disarmament research institute has warned, as Pakistan cautioned that advances in AI and the life and chemical sciences are creating new and unpredictable risks.
Robin Geiss, director of the UN Institute for Disarmament Research (UNIDIR), put a stark question to the international community during a briefing at UN headquarters: “Are we going to let machines target and kill human beings?”
He warned that military AI is increasingly “on track to crowding out human decision-making”, including in the nuclear domain. AI systems are already used for intelligence analysis, target identification, drone operations and cyberattacks. Their expanding role is speeding up battlefield operations, shrinking the time between identifying and striking targets, and letting forces engage unprecedented numbers of targets each day.
Why AI oversight in warfare matters now
Mr Geiss cautioned that the opacity of many AI systems raises the danger of errors, miscalculations and unintended escalation, and that easier-to-use tools could let non-state actors run sophisticated military operations.
He acknowledged that geopolitical competition hampers regulation, but said concern about the risks now crosses political divides. In September, the United States proposed an AI “notification mechanism” with China for alerting each other to AI incidents threatening national security, a sign even rivals see the danger.
He also noted growing support for UN Secretary General Antonio Guterres’s call for a legally binding treaty on lethal autonomous weapon systems, and for the view that technology must not cross a “moral red line”.
New UN forum on military AI
UNIDIR will pursue that cooperation through its Centre of Excellence on AI, Peace and Security, which launches a new forum on Thursday on AI, international peace and security, and the military domain — the first and only such forum in the UN system, Mr Geiss said.
It will develop voluntary guidelines for industry, a common vocabulary on concepts like autonomy and safety, and measures against misuse by non-state armed groups.
The push comes as governments race to fold AI into early-warning systems and targeting, blurring the point where human oversight still counts. For nuclear-armed Pakistan, living with compressed decision timelines in South Asia, that question is far from academic, as Islamabad’s strategists noted this week.
Pakistan stresses cooperation at the UN
Separately, Pakistan highlighted the convergence of AI, synthetic biology and the life and chemical sciences at the UN General Assembly’s First Committee.
“Science is moving faster than governance,” Counsellor Syed Atif Raza told the committee, warning that emerging technologies were creating novel risks.
The intervention fits Pakistan’s consistent line. Last month, Foreign Minister Ishaq Dar told the Security Council that AI must not become “another facet of the ongoing arms race”, urging guardrails and governance frameworks. Pakistan has pressed for restraint and confidence-building in the military domain, a theme echoed in recent warnings from its security establishment.
The debate unfolds as states pour resources into new defence arrangements across the region. Mr Geiss’s question, whether machines will be allowed to target and kill human beings, will now be tested in the forum UNIDIR launches this week.
Frequently asked questions
What did the UN warn about AI in warfare?
UNIDIR director Robin Geiss warned that military AI is on track to crowd out human decision-making, including in the nuclear domain, and asked whether the world will let machines target and kill human beings.
What is UNIDIR’s new forum on military AI?
It is a new forum under UNIDIR’s Centre of Excellence on AI, Peace and Security, launching this week, aimed at voluntary industry guidelines, a common security vocabulary on autonomy and safety, and curbing misuse by non-state armed groups.
What did Pakistan say about AI risks at the UN?
Counsellor Syed Atif Raza told the UNGA First Committee that science is moving faster than governance, warning that the convergence of AI, synthetic biology and life and chemical sciences is creating novel risks.
Is there a treaty on lethal autonomous weapons?
Not yet. UN Secretary General Antonio Guterres has called for a legally binding treaty on lethal autonomous weapon systems, and support for that call is growing, according to UNIDIR.
