Notes on Leo XIV’s first encyclical, Magnifica Humanitas, and what its framing means for those of us working in European AI.
When Pope Leo XIV said, at the launch of his first encyclical earlier this month, that “artificial intelligence needs to be disarmed,” he chose the word carefully and said so himself. Magnifica Humanitas, published on 15 May 2026 and subtitled On Safeguarding the Human Person in the Time of Artificial Intelligence, is the first major magisterial document to engage directly with AI. The launch address that accompanied it placed the word “disarm” alongside the Church’s long history of work on nuclear disarmament — a deliberate parallel that asks a specific question about scale, power and accountability.
I have spent the last few weeks reading the document slowly. I want to set out why I think its framing matters, particularly for those of us working in or around European AI policy and practice — and why “disarmament” is not, on closer reading, the anti-technology gesture it might first appear to be.
What “disarm” actually means
The argument in Magnifica Humanitas is not against AI. It is against AI as an instrument of domination, exclusion and death. Leo XIV is explicit that technology is a deeply human reality, bound up with human freedom and creativity, and that each historical phase of technological progress has improved the conditions of human life. What he warns against is not the tool but the unaccompanied tool — a technology that has outpaced the moral, political and institutional frameworks that should be holding it to account.
The two examples he named at the launch are precisely the cases European regulators have been wrestling with for years: autonomous weapons systems beyond meaningful human control, and algorithms that gate access to healthcare, employment and security on the basis of data tainted by bias. Neither is a hypothetical. Both are operating now, in production, in systems that affect millions of people. The encyclical’s contribution is not to discover these problems but to give them a name that travels — and “disarmament” travels.
Why this lands in a European register
There is a striking alignment between the document’s framing and the European conversation about AI. Both place the dignity of the person at the centre, rather than the efficiency of the system. Both treat technology as something whose moral character depends on its design, financing and deployment rather than something neutral that simply “is.” Both insist that the public has a legitimate interest in the way these systems are built and governed.
One passage in particular stopped me. The encyclical observes that the main engines of today’s technological progress are private and often transnational, and command resources beyond the reach of many national governments. The principle of subsidiarity — the idea that decisions should be taken as close as possible to those affected by them — is reframed here for a world in which the entities operating at the apex are no longer States but a small number of platforms. That is, I think, the most lucid sentence written about AI governance this year. It captures something that European policymakers have intuited for some time but have struggled to articulate without sounding either alarmist or naïve.
The document also moves data, algorithms, patents and digital infrastructure into the category of goods subject to “universal destination” — the principle that the earth’s goods are intended for the use of all. That is a significant theological and political claim. It says that the wealth generated by knowledge and technology cannot legitimately remain concentrated without producing a new kind of imbalance between the included and the excluded. Whether or not one shares the document’s metaphysics, the practical implication maps directly onto the conversation we are having in Europe about open data, interoperability, sovereign infrastructure and the conditions of access to AI capability.
Babel and Nehemiah
The encyclical’s governing image is the contrast between two biblical building projects. The Tower of Babel: a single project, a single language, a single direction, built to “make a name” with no reference to anything beyond itself. And the rebuilding of Jerusalem under Nehemiah: a community convened, the work distributed, each group responsible for its own section of the wall, the architecture emerging from coordination rather than imposition.
Leo XIV’s argument is that the real question about AI is not yes or no, but which of these two projects we are working on. That framing is more useful than the binary that usually structures the debate. It refuses both technophilia and technophobia and replaces them with a question about form: what kind of project is this, who is participating in it, and what makes it stable?
For those of us in Europe trying to articulate what a different model of AI development might look like — one that is neither a defensive imitation of American hyperscaler economics nor a defeated submission to it — the Nehemiah image is genuinely useful. It is a picture of distributed work under shared coordination, with the architecture emerging from the contributions of many rather than from the will of one. That is recognisably the shape of what serious European AI policy is trying to build, even when it does not yet have the vocabulary to describe itself.
A document that takes time seriously
What I appreciated most, reading Magnifica Humanitas, was the absence of urgency-theatre. The document does not perform the fear that so much AI commentary performs, and it does not chase the news cycle. It locates the present moment within a hundred-and-thirty-five-year lineage of social thought, names its predecessors carefully, and treats the question of AI as part of a much longer conversation about work, dignity, the common good and how human beings build things together.
That is a register we are very short of right now. The AI conversation is dominated by voices speaking either from a position of imminent salvation or imminent collapse, and very little of it is durable enough to be re-read in five years. Magnifica Humanitas is durable. It will be re-read.
The full text is available at vatican.va in several languages. The launch address is worth watching alongside the document — the choice of the word “disarmed” lands differently when spoken than when read. Whatever your relationship to the Church, the document repays the time.
