Pope Leo XIV Releases 'Magnifica Humanitas' — 42,300-Word Encyclical Calls on Governments to Disarm AI

Pope Leo XIV published “Magnifica Humanitas” on May 25, 2026 — a 42,300-word papal encyclical dedicated entirely to the ethics and governance of artificial intelligence. It is the first papal document in history to make AI its singular subject, and it arrives at a moment when regulatory frameworks from Brussels to Beijing are still being written.

The document does not ban AI or call for a moratorium. It calls for what the pope terms “disarmament” — a deliberate, multilateral slowdown in deploying AI in high-stakes domains (weapons systems, judicial decisions, social credit scoring) without binding international oversight. The language echoes nuclear non-proliferation rhetoric, not accidental framing.

The four risks the encyclical names

The pope structures his warnings around four categories:

1. Concentration of power: AI capabilities concentrate in a small number of corporations and states. The encyclical argues this creates asymmetric leverage over democratic processes, labor markets, and information flows — a condition incompatible with human dignity in Catholic social teaching.

2. Distortion of truth: Large language models and synthetic media systems are explicitly named as tools capable of manufacturing consent at scale. The document quotes researchers from MIT and Oxford who have measured truth-decay in AI-heavy information environments.

3. Labor displacement: The encyclical does not argue against automation in principle but insists that the displacement of workers without social safety nets violates the dignity of labor. It calls for national governments to tie AI deployment permits to worker transition plans.

4. Militarization: Autonomous weapons systems receive the longest treatment in the document. Leo XIV argues that delegating lethal force to algorithmic decision-making violates the principle that moral responsibility must attach to a person who can be held accountable.

Christopher Olah at the Vatican

The presence of Christopher Olah — Anthropic co-founder and one of the most cited researchers in AI interpretability — at the encyclical’s presentation was intentional and symbolic. Olah did not co-author the document, but he participated in the Vatican’s year-long consultation process that informed it. Anthropic is the only major AI lab to publicly engage with the Vatican’s AI ethics process.

The symbolism cuts both ways. Critics immediately noted that Anthropic builds and commercializes the same AI systems the encyclical warns about. Supporters pointed to the company’s Alignment Science team and its focus on mechanistic interpretability as evidence of genuine engagement with the risks named in “Magnifica Humanitas.”

Practical weight of the encyclical

Papal encyclicals are not law. They carry no enforcement mechanism. But they carry significant moral authority across 1.4 billion Catholics and a demonstrated track record of shifting policy debates. “Laudato Si’” on climate change in 2015 is the modern reference point: it directly influenced national policy positions at COP21 in Paris.

The EU AI Act’s governance framework is already closer to the encyclical’s prescriptions than any other major regulatory regime. The document may give political cover to lawmakers in traditionally Catholic countries — Brazil, Italy, the Philippines, Poland — who want to push for stricter AI governance domestically or in international forums.

Whether “Magnifica Humanitas” moves AI policy meaningfully or remains a moral statement without legal teeth will depend on who chooses to cite it and in what chambers.