Cardinal Michael Czerny says responsibility for violence involving AI must remain identifiable and human, with lethal-force decisions made by people who have time to exercise judgement. His three tests, set out in Sofia, put accountability, deliberation and civilian protection at the centre of the debate over AI in combat.
Watch Desk analysis
What happened
Speaking on 17 November at a gathering co-hosted by the Bulgarian Academy of Sciences and the Apostolic Nunciature in Bulgaria, Czerny presented Pope Leo XIV’s encyclical Magnifica humanitas. He said the chain of responsibility for AI systems used in warfare should be identifiable and verifiable, from those who design and train them to those who authorise and use them.
Czerny also said decisions about lethal force must stay in human hands, and that decision-makers need time to exercise judgement. Speed and efficiency, he argued, cannot determine an irreversible decision. His third test is that civilians must be distinguished from combatants and protected, along with the infrastructure they need to survive. These are Czerny’s account of the encyclical’s criteria, as reported by Vatican News.
Why it matters
Those tests turn the broad call for “human responsibility” into questions that can be put to governments and developers: who is accountable along the system’s chain, who makes the final decision, and whether they have enough time to make it. That is a more useful starting point than treating a human somewhere in the process as a magic safety charm.
The protection of civilians and essential infrastructure also makes clear that the argument is not only about who presses a button. It is about what systems are used to do, and what safeguards surround decisions with irreversible consequences.
Our read
Czerny’s strongest point is the insistence on a traceable chain of responsibility. If a decision cannot be connected to identifiable people, accountability risks disappearing into the machinery. The three tests are a framework for scrutiny, not evidence that any particular military system meets them. Still, they give the debate something concrete to work with, rather than another round of grand statements about keeping people “in the loop”.
What to watch
- Whether governments set out who is accountable at each stage of AI-enabled military systems.
- How decision-makers’ time and authority are protected when lethal force is considered.
- Whether civilian protection and essential infrastructure feature in rules for deploying these systems.
Discussion spark: Should AI-assisted military systems be barred from decisions about lethal force unless every step has a clearly identifiable human decision-maker, or can meaningful accountability work another way?
Sources and evidence
- Source update (18 November 2026, 10:18 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.