Notes on Pope Leo XIV’s encyclical on AI
25 may 2026
Published by the Vatican this morning: Magnifica Humanitas of His Holiness Pope Leo XIV on the Protection of the Human Person in the Age of Artificial Intelligence. this is one very interesting document. This is the clearest writing I have seen on the ethics of integrating AI into modern society.
Pope Leo XIV chose the name Leo in honor of Pope Leo XIII, known for his 1891 rerum novarum Encyclopaedia on “The Rights and Duties of Capital and Labour”.
This story on Vatican News further makes clear the significance of that decision:
Meeting with the College of Cardinals for the first formal meeting after his election, Pope Leo XIV explained part of his reason for choosing his papal name. “There are different reasons,” he said, before explaining that he chose the name Leo “mainly because Pope Leo XIII in his historical encyclical rerum novarum “Addressed the social question in the context of the first great industrial revolution.”
In “Our Own Time,” the Church offers everyone the treasure of her social teaching in response to another industrial revolution and developments in the field of artificial intelligence that pose new challenges to the defense of human dignity, justice, and labor.
And now we get Pope Leo XIV’s own encyclical on the AI revolution. There is a lot here, but the writing style is very accessible, including to non-Catholics.
some of my glimpses
(I listened to most of the encyclopedias while on walks with my dog. I was trying the ElevenReader iPhone app for the first time. It worked great: I pasted a URL into the document and it read it to me in a very high-quality voice, highlighting each paragraph as I read it.)
Here are some of my highlights. In each case below emphasis Is mine.
Here is a useful description of the interpretability problem for LLM in Section 98:
First, given the remarkable pace at which these systems are developing, any statement regarding AI is in danger of quickly becoming outdated. Second, we all, including those who design them, have only a limited understanding of their actual functioning. In fact, Current AI systems are more “cultivated” than “built”, as developers do not directly design every detail, but rather create a framework within which intelligence “grows.” As a result, fundamental scientific aspects – such as the internal representations and computational processes of these systems – currently remain unknown.
I liked the description of the relationship between development and dignity in section 83:
Development is both a duty and a right for individuals as well as nations. Minimum conditions are required to enable every person and people to flourish in accordance with their dignity, without being placed in a state of dependency or without access to essential goods. Development is truly humane when it puts people at the center rather than wealth accumulation, and when it concerns people as well as individuals. Justice demands recognition of the rights of society and the rights of people, and also includes responsibility towards future generations. Development is not truly humane if it increases consumption for some while shifting costs and burdens onto others, or forces entire sectors into subordinate roles, preventing them from realizing their full potential..
Cultural prejudices and sycophancy are mentioned in Section 100:
In personal use, three aspects in particular need to be carefully considered: ease of obtaining results, impression of fairness, and simulation of human communication. The speed and ease with which information, complex analysis, media content and practical assistance can be accessed undoubtedly makes life easier. Yet they may also encourage excessive reliance and the search for ready-made answers, and undermine individual creativity and judgment. The apparent objectivity of the feedback and suggestions provided by these systems may lead us to overlook the fact that they reflect the cultural assumptions of those who designed and trained them, with all their strengths and limitations.. Artificial imitations of positive human communication – words of advice, sympathy, friendship and even love – can be attractive and sometimes actually helpful. However, for less discerning users, this may also be misleading, creating an illusion of connection with the actual individual subject.. When words are imitated, they do not create actual connections, but only form them. Artificial imitation of care or support can be especially risky when it enters into contexts where real relationships and emotional bonds are lacking.
101 Touch on Environmental Impact:
Current AI systems require huge amounts of energy and water, which significantly impacts carbon dioxide emissions, and places huge demands on natural resources. As their complexity increases, especially in the case of large language models, the need for computing power and storage capacity also increases, requiring extensive networks of machines, cables, data centers, and energy-intensive infrastructure.. For this reason, it is necessary to develop more sustainable technological solutions that reduce environmental impact and help protect our common home.
102 covers the risks of algorithmic systems making decisions that affect people’s lives without “compassion, mercy, forgiveness”:
The use of AI is never a purely technical matter: When it penetrates the processes affecting people’s lives, it touches rights, opportunities, status and freedom. Important and sensitive decisions – related to employment, credit, access to public services or even a person’s reputation – The risk is being handed over to completely automated systems that do not know “compassion, kindness, forgiveness and above all, hope that people are capable of changing”. and hence may give rise to new forms of exclusion.
105 The need for human accountability in how these systems are implemented emphasizes:
For AI to respect human dignity and truly serve the common good, responsibility must be clearly defined at every stage: From the people who design and develop these systems to the people who use them and rely on them to make sound decisions. However, in many cases, the internal processes leading to the outcome remain opaque, making it difficult to assign responsibility and correct errors. This is where accountability becomes important: the possibility of identifying who should “account” for decisions, justifying them, monitoring them, and, when necessary, challenging them and addressing any harm caused.
And 108 touches on the way AI amplifies the power of those with resources:
In fact, as with every major technological change, AI amplifies the power of those who already have economic resources, expertise, and access to data. In light of the common good and the universal destination of goods, this raises serious concerns, as small but highly influential groups can shape information and consumption patterns, influence democratic processes and steer economic dynamics to their advantage, thereby undermining social justice and solidarity among people. For this reason, it is essential that the use of AI, especially when it touches public goods and fundamental rights, is guided by clear norms and effective oversight, based on participation and subsidiarity.
The same section explicitly calls data something that should be thought of as a public good:
[…] moreover, Ownership of data cannot be left solely in private hands But should be properly regulated. Data is the product of many contributors and should not be treated as something to be sold or handed over to a select few. It is necessary to think creatively to manage data in a spirit of participation, as a common or shared good, as St. John Paul II already suggested in relation to collective goods.
Given that Palantir is named after a lord of the rings context, I can’t help but wonder whether JRR Tolkien’s quote return of the king (Section 213) Was the Pope throwing a little shade at Peter Thiel.
Twentieth-century Catholic author JRR Tolkien, in the words of the protagonist of one of his novels, described our responsibility this way: “It is not our part to overcome all the tides of the world, but to do all that is in us to help the passing of the years, to root out evil in the regions we know, that they who live after may have a clear earth to till.” The civilization of love will arise not from a single or showy gesture, but from the sum of small and determined acts of loyalty that serve as a bulwark against dehumanization. For this reason, it is worthwhile to consider some aspects of how we, each in our own way, can contribute to the building of a civilization of love.
Another prediction for 2026 below
On January 6th of this year I joined Oxide and Friends’ 2026 Predictions Podcast episode to talk about predictions for 2026, 2029, and 2032. I wrote my article here, taking the view that they weren’t nearly ambitious enough – it’s already undisputed that LLMs write good code, we’ve made huge progress in sandboxing and the New Zealand Kakapo have really excellent breeding season.
There is a section of this episode that I didn’t bother to include in my article, but I can’t resist providing it here as a lightly edited transcript:
Brian Cantrill: 37:13
I think AI has created some real public perception problems for itself. And I think this year you will have, from one of the leading model companies, a white paper explaining how the proliferation of AI will mean prosperity for everyone. They may be trying to make some economic argument – because this will be a 2026 election issue, how we think about these things and how they are regulated and it is a big mess. There is more heat than light in this debate.
Simon Willison: 38:05
I would like to tag something on that: I think it only works if they can wash it through existing trusted experts. Sam Altman and Dario are constantly publishing essays on this topic and no one believes a word they say. Get Barack Obama’s signature on one of these term papers Perhaps You have something that people can start to trust a little bit.
Adam Leventhal: 38:27
Otherwise, it’s just like “leaded gas is good for you”, says Exxon.
Brian Cantrill: 38:31
I mean yes. God. Obama… Let’s go with that, it’s great because if it’s like Bill Clinton everyone will roll their eyes, so it has to be someone who has real credibility to say this will be broad-based… I would say if they asked that person to do it, it would turn out that it’s also a little skewed.
Simon Willison: 38:57
How about the Pope?
Brian Cantrill: 39:01
The Pope is very interested in this thing! This is a great prophecy. We’ve hit pay dirt. The Pope is considering LLMs and their economic impact on the world.
Simon, if the Pope believes this will be an economic disaster then I’m giving you full credit.
My prediction here seems much less plausible given the Leo XIV/Leo XIII relationship, which I had no idea about when we recorded the episode!
<a href