Pope Leo XIV Understands AI as Power. He’s Right. Now He Must Name the Actors Who Wield It

Pope Leo XIV’s first encyclical, Magnifica Humanitas, is dedicated entirely to AI and its consequences for humanity. Released on May 25 to coincide with the 135th anniversary of Leo XIII’s Rerum Novarum on capital and labor, it goes much deeper than the vast majority of corporate statements on the ethics of AI, and is much more honest than many government white papers that avoid challenging Big Tech.

The encyclical is right in the essentials: AI is not just another technology, a neutral tool that we can evaluate solely for its efficiency, its precision or its ability to reduce costs. It is an infrastructure of power. It decides what we see, what we read, which jobs disappear, which decisions are automated, what forms of surveillance are normalized, and what parts of shared reality end up turned into noise, polarization, or spectacle. As Wired summarizes, the Pope understands AI as an invisible layer that cuts across work, information and collective decisions. So far, so good.

The problem is that, when the analysis should become an accusation, no names are mentioned. The Pope highlights the concentration of power, but does not name those who concentrate it. He talks about platforms, but avoids naming them. He discusses market logics, but does not identify the companies that have turned these logics into a form of private government over our lives. And that is not a theological problem, it is a political decision. It’s an encyclical that talks in generalities and the abstract.

But reality is not abstract. While the pope warns against the culture of power and even calls for AI to be “disarmed”, the US Department of Justice has evidence that Google has used its dominant position to control the search and video sharing market for a decade, and now intends to use Gemini to do the same with AI. We are not talking about a “worrying trend”, but about a monopoly identified by the courts, of exclusivity contracts, of data, of distribution, of the very architecture of access to information. The pontifical document sees the building, but it avoids reading the sign on the door.

The same goes for Meta. Wired’s story on protests by subcontracted content moderators that label data to train Meta’s AI protesting in Dublin who are massively underpaid, we’re not reading about labor mismanagement. This is what happens when AI is deployed within real value chains: invisible, outsourced, psychologically expensive work, presented as if it were technological magic, and the people who do it are simply discarded when no longer needed. Where is the dignity of work, where exactly in the chain is the “centrality of the human being?”

The part of the encyclical about the information ecosystem is the most interesting. Here the Pope is absolutely right: the problem is not only misinformation, hoaxes or AI slop misleading the unwary for a few hours. The problem is much deeper: it is the algorithmic reconstruction of the collective imagination. A system that optimizes for engagement, indignation, permanence and emotional reaction not only distributes content: it defines what a society considers visible, debatable or possible. A study published in PNAS Nexus showed that engagement-based ranking algorithms amplify emotionally charged and hostile content toward political adversaries; another study collected by El País pointed out that the order in which political messages are displayed affects polarization. In practice, the architecture of public conversation is no longer designed by parliaments, publishers or citizens, but by opaque optimization models that turn our most primal impulses into advertising inventory.

And here the encyclical is valuable because it understands something that many regulators have yet to grasp: fact-checking does not fix a system built to reward the opposite of deliberation. Fact-checking may be necessary, but it is desperately insufficient when the entire infrastructure is designed to appeal to emotions. Democracy is not destroyed by publishing lies, but when the common space in which we should distinguish between truth, lies, interest and manipulation has been redesigned by private companies that answer to advertisers, shareholders and internal metrics, not to citizens.

Where the document falls shortest is on work. Pope Leo defends the dignity of labor, but from a conception of employment that sounds increasingly out of date. The big question is no longer whether AI will destroy jobs, but who captures the value of that destruction, that substitution or that reorganization. When a company automates a process, reduces headcount, increases productivity and improves margins, where does that value go? To the workers? To their communities? Into taxes capable of financing transition, training or income? As we know, most of it goes to the markets. McKinsey formulates it from the language of business strategy: AI is not simply a productivity revolution, but a competitive reset, a sinister reconfiguration of business models, market structures and profit pools.

That is why South Korea’s proposal for a citizen dividend from AI profits is so interesting. It may be imperfect, debatable, or premature, but at least it puts the question in the right place: if AI systems are trained on data, labor, infrastructure, collective knowledge, and socially constructed markets, why should the return be concentrated almost exclusively on a few companies and their shareholders? Why do we accept without batting an eye that automation is presented as progress when its benefits are privatized and its costs are socialized?

The Vatican has enough tradition to go much further. Rerum Novarum did not limit itself to bemoaning “the excesses of industrialization.” It pointed out abuses, spoke of wages, property, employers, workers, associations and concrete conditions. It understood that social doctrine could not remain a pious appeal to the common good while factories, landowners and industrial capitalism redefined the lives of millions of people. Magnifica Humanitas wants to be the Rerum Novarum of AI, and in many ways it succeeds: it understands the change of era, recognizes the magnitude of the challenge and refuses to fall into technophile naivety. But just when it should transform the diagnosis into a concrete institutional critique, it takes refuge in abstraction.

Obviously, this is not about asking the Pope to write a blacklist of tech companies, or to turn an encyclical into an antitrust lawsuit. Encyclics are designed to be current and readable for decades or centuries, so mentioning specific companies makes no sense. But it does mean understanding that in 2026 talking about AI without indicating that companies like Google, Microsoft, Meta, Amazon, OpenAI, Nvidia, Apple or Anthropic is like talking about the Industrial Revolution without mentioning factories, mines, railways, employers or unions. Not a specific mention, but an indication of which types of actors are good and which are bad. Technology does not fall from the sky. Technology has owners, incentives, contracts, supply chains, business models, and regulatory capture strategies. When all this disappears from language, the possibility of acting on it also disappears.

None of this diminishes the importance of the document. In a world where the ethics of AI are too often being produced by the very companies that benefit from its absence, for an institution with global moral authority to say that technology should serve humanity rather than corporations is highly relevant. Making the central issue power, is also important. Discussing communication, education, work, war, democracy and dignity in the same framework is much more sophisticated than the usual litany of “ethical principles” designed to adorn corporate presentations.

Nevertheless, it falls short. The next letter, encyclical or next synodal document will have to do what it avoids: point out structures, identify actors and propose concrete mechanisms rather than make a generic appeal to multilateralism in a world in which multilateralism cannot compete with lobbies. We don’t need solemn invocations of shared responsibility while value capture remains brutally asymmetrical. No more impeccable diagnoses undermined by timidity in accusation.

The Pope shows that he knows what the problem is. He understands that AI is not a tool, but a new architecture of power. He has seen that it affects truth, work, democracy, war and freedom. He has even found a powerful formulation: disarming AI. But disarming something requires knowing who has the weapons, who manufactures them, who sells them, who benefits from them and who pays the consequences of their use.

And there, precisely there, Magnifica Humanitas falters. The Pope’s diagnosis is lucid, comes from the moral high ground, recovers a valuable social tradition and asks necessary questions. But he avoids pointing the finger at the evidently guilty parties.

The Pope has made the diagnosis. Now he must find the courage to name the source of the problem.

—

This post was previously published on Enrique Dans’ blog.

***