Joint declaration signed in Geneva against use of AI in warfare

By Valerio Palombaro

Warfare “accelerated” by artificial intelligence (AI) is increasingly becoming “a means of enabling killing rapidly and at scale.” At the same time, there are currently no “technical or procedural solutions capable of preventing the lethal and devastating consequences” arising from the challenges these developments pose to international law.

Against this troubling backdrop, the World Council of Churches (WCC), together with 225 other signatories—including NGOs, associations, experts, and representatives of technology companies—has endorsed a joint declaration opposing the use of AI in warfare.

UN meeting in Geneva

The declaration explicitly calls on companies developing AI technologies and on governments to “cease supplying” artificial intelligence systems intended for use in the “military kill chain” and to “take all necessary measures to ensure that any other AI systems they provide do not cause or contribute to violations of international humanitarian law and international human rights law.”

The declaration was signed during a meeting taking place in Geneva from 15 to 17 June, organised by the United Nations Office for Disarmament Affairs, on the theme of AI in the military domain and its implications for international peace and security.

The gathering comes only weeks after Pope Leo XIV, in his Encyclical Magnifica Humanitas, called for the world to “disarm AI.”

The risk of “diluting” human responsibility

According to the declaration, all companies—including those contracting with government military agencies—“must take every possible measure to ensure that their products and services do not cause, contribute to, or become directly linked to human rights abuses and international crimes.”

Furthermore, “where companies are unable to prevent or meaningfully mitigate such risks, they should not enter into or fulfil such contracts.”

Citing media reports and official Pentagon statements, the declaration notes that “the rapid generation of targets through AI tools has enabled an increase in the speed, scale, intensity, and destructive force of U.S. strikes against Iran.” Similar assessments, it adds, can also be made regarding systems employed by the Israeli armed forces.

Such technologies, the declaration argues, dilute “human responsibility in life-and-death decisions” and “may help conceal international crimes behind a veneer of apparent algorithmic objectivity while simultaneously evading accountability.”

An inhumane drift

“The real-world deployment of AI indicates that it is in fact facilitating more violent, dehumanising, and destructive methods of warfare,” the 226 signatories state. Among them are international organisations such as Amnesty International, as well as numerous local groups.

“We are particularly concerned,” the declaration continues, “that the use of large language models (LLMs) for target generation and prioritisation is pushing military actors toward a form of warfare in which the fundamental principles of international humanitarian law—including distinction, proportionality, and precaution—are not, and likely cannot be, adequately respected.”

Given the speed and scale of these technologies, together with the unreliability, bias, and often unlawfully sourced nature of the input data, the signatories warn that such systems risk facilitating human rights violations, crimes against humanity, and war crimes.

Advertisement

Moreover, the opacity surrounding their use seriously undermines the possibility of assigning moral or legal responsibility when mistakes occur.

Meaningful human control

The declaration further argues that, even when AI systems used in targeting do not make the final decision to kill, they risk becoming mechanisms for the automatic approval of large-scale killing because they rely on a false sense of objectivity and may displace human responsibility and due diligence.

As a result, they can contribute to accelerating and streamlining mass killings.

These systems also automate dehumanisation by “reducing matters of life and death to a simple chat prompt.” The decision to take a human life, the signatories stress, carries profound moral and legal weight and must never be reduced to the acceptance or rejection of recommendations generated by AI systems.

When militaries rely on AI to identify targets at such speed and scale that human review becomes little more than a formality, “mass atrocities can, and often will, occur,” in direct violation of the precautionary principle enshrined in international humanitarian law.

Greater accountability and transparency

The declaration signed in Geneva calls on technology companies to refrain from “entering into or executing contracts with military agencies or armed groups implicated in potential violations of international law, including human rights abuses and atrocity crimes.”

Governments, meanwhile, are urged to “halt the use of AI tools, including large language models, in military targeting operations and ensure compliance with international humanitarian law and international human rights law.”

They are also called upon to “ensure transparency regarding the current use of AI in the conduct of hostilities.”