Back AI in conflict: the time to disarm AI is now

Commissioner's Shout Out
AI in conflict: the time to disarm AI is now

Imagine a dystopian world where algorithmic profiling turns military target selection into an automated "factory of targets," tracking individuals to their family homes before approving airstrikes.

This is no longer dystopia. The decisions that determine who lives and who dies in war are increasingly being shaped by machines. Artificial intelligence (AI) is rapidly being integrated into military operations, from intelligence gathering and target identification, to weapons systems and decision on the use of force. Autonomous weapon systems, including AI-enabled ones capable of selecting and engaging targets without human intervention once activated, are already being developed and deployed.

Next week offers states a chance to act. From November 16-20, governments will meet in Geneva to consider opening negotiations on a treaty to regulate autonomous weapons. The UN gathering provides an opportunity to enshrine a key principle in law: humans must retain meaningful control over the use of force and remain accountable for it. States must seize this moment to disarm AI.

A test for human control

The question is not whether a military system uses AI. Used responsibly, unmanned and AI-enabled systems can help protect civilians or reduce soldier casualties. The issue is whether humans retain meaningful human oversight and whether these technologies can be used in compliance with international law. AI systems can be opaque and unpredictable, drawing on outdated or biased data to produce outcomes that cannot be adequately anticipated. Errors have already occurred, with devastating results. Furthermore, the speed of AI-enabled operations can leave too little time to verify targets or assess the legality of engaging them. Human control risks becoming a rubber stamp when operators defer to machines they perceive as more authoritative or precise than themselves or when they lose control of autonomous decision-making altogether. These dynamics heighten the risk of unlawful attacks, threaten the right to life and erode accountability for violations.

The risks do not stop at the battlefield. Technologies developed for war migrate to civilian settings, such as border control or the policing of assemblies, or fall into the hands of non-state actors committing unlawful acts. AI is weaponised to manipulate the information environment and disrupt civilian infrastructure.

Closing a regulatory gap

Regulation has failed to keep pace with these realities.

While international human rights law and international humanitarian law apply and must be upheld, their interpretation does not settle every question these technologies raise. Existing frameworks, including the EU AI Act and the Council of Europe Framework Convention on Artificial Intelligence, exclude military uses from their scope.

For more than a decade, the United Nations, the International Committee of the Red Cross and civil society organisations have worked to address these dangers, reflecting a growing consensus that existing legal guardrails are insufficient. Important processes are now underway at the UN, both within the CCW* and at the General Assembly, where states are discussing the regulation of autonomous weapons and the wider consequences of military AI for peace and security.

Preserving human dignity

As the race to militarise AI accelerates, decisive action is imperative. States should act swiftly by supporting urgent, time-bound negotiations for a global, legally binding instrument on autonomous weapons, and by advancing regulation of other military uses of AI.  

The private sector also bears a responsibility. Companies that design, develop or supply these systems should respect human rights in line with the UN Guiding Principles on Business and Human Rights. They must conduct human rights due diligence, mitigate risks, and embed human rights and humanitarian law standards into the design and deployment of their products.

Decisions over life and death must remain anchored in human judgement, responsibility and accountability. The use of force must be governed by people acting under clear rules and respecting human rights – never delegated to machines. Humanity demands no less.


*Convention on Certain Conventional Weapons


 Read more about AI and human rights

Strasbourg 13 November 2026
  • Diminuer la taille du texte
  • Augmenter la taille du texte
  • Imprimer la page