The Pope’s 42,000-word verdict on AI

The Pope Just Wrote 40 Pages About AI. Silicon Valley Should Read Every Word.

You know something has officially crossed over from “tech news” to “civilization news” when the Pope writes a letter about it.

Not a tweet. Not a passing comment at a press conference. An encyclical. The most solemn, most considered form of papal teaching there is. Only a handful of these are issued in a typical papacy. They’re reserved for things that keep the Pope up at night. War. Poverty. The family. The dignity of human labor.

And now: artificial intelligence.

On Tuesday, Pope Leo XIV released Magnifica Humanitas — “The Greatness of Humanity” — addressed to the Church’s 1.4 billion members worldwide. But the intended audience is much wider. The encyclical is a warning, a call to action, and, in its quiet way, a political intervention. Its argument, boiled down to a single sentence, is this: a moral AI means nothing if that morality is determined by a few private companies operating without accountability, and the world had better “disarm” this technology before it dominates humanity.

To understand why this matters, you have to understand what the Pope is not doing. He’s not condemning AI. He’s not telling Catholics to avoid chatbots or cancel their cloud subscriptions. He’s not calling for a Luddite smash-and-burn. In fact, Magnifica Humanitas explicitly acknowledges the potential benefits of AI in medicine, education, translation, and scientific research.

But Leo is drawing a line. And he’s chosen an unusual ally to help him hold it: Anthropic’s Christopher Olah, who stood with the Pope during the encyclical’s presentation and delivered a remarkably candid statement about the incentives inside frontier AI labs.

This is not your grandfather’s Vatican.

The Core Argument: AI Is Never Neutral

The encyclical is dense—about 40 pages in its English translation—but the structure is clear. Leo builds his case in three movements.

First: AI is not neutral.

This sounds obvious to people who’ve worked in tech for more than five minutes, but it’s surprisingly controversial in public discourse. How many times have you heard “AI is just a tool, it depends how you use it”? Leo rejects that flatly. A tool designed by people, trained on human-generated data, deployed within corporate structures, and optimized for metrics chosen by executives is never neutral. It carries the values, biases, and incentives of its creators.

He writes: “Every algorithm is a confession of priorities. It reveals what its makers value, what they measure, and what they are willing to overlook.”

That’s not Luddism. That’s systems thinking. And it’s the foundation for everything else.

Second: The drivers of AI are private, transnational companies that already surpass the capacity of many governments.

Leo doesn’t name names—encyclicals are not hit pieces—but he doesn’t need to. Everyone knows who he means. The handful of companies that control frontier models have valuations larger than the GDP of most nations. They make decisions about safety, deployment, and access that affect billions of people. And they do so with minimal democratic oversight.

“It is a strange silence,” he writes, “when the most powerful artifacts of our age are governed not by legislatures but by product managers.”

That line will sting. It’s meant to.

Third: The goal is not to stop AI but to disarm it.

Here Leo uses deliberately martial language. He calls for AI to be made “human-friendly” and freed from monopolistic control. He warns that without intervention, AI will reduce people from ends to means—from subjects of their own lives to cogs in an efficiency machine.

This is the most traditionally Catholic part of the encyclical. It draws on a century of social teaching about labor, capital, and the dignity of the human person. Leo is essentially arguing that AI is the Industrial Revolution of our time, but with a crucial difference: the Industrial Revolution centralized physical capital (factories, railroads, mines), while AI centralizes cognitive capital (reasoning, planning, decision-making). That’s more intimate. More dangerous.

The Practical Ask: Four Pillars

Leo doesn’t just preach. He proposes a concrete framework: four pillars for governing AI.

Robust legal frameworks – Not voluntary codes of conduct. Not “we promise to be good.” Laws. With enforcement. With consequences.

Independent oversight – Regulators who are not captured by the industries they regulate, with technical expertise and teeth.

Informed users – Public education about how AI works, what it can and cannot do, and who profits from its use.

A political system that does not abdicate its responsibility – This is the dagger. Leo is calling out governments that have outsourced AI governance to the companies themselves. He’s saying: do your job.

These are not radical demands. They’re the same things consumer protection advocates, AI safety researchers, and even some industry insiders have been saying for years. The difference is that now the Pope is saying them. And when the Pope says something, bishops read it in parishes. Catholic universities hold symposia. Policy advisors in majority-Catholic countries take notes.

The Red Line: No Algorithms on the Trigger

The most concrete, least ambiguous section of Magnifica Humanitas concerns lethal autonomous weapons.

“No algorithm can make war morally acceptable,” Leo writes. “The decision to take a human life, even in the just defense of others, requires a human conscience, a human judgment, a human soul. To delegate this to a machine is not only dangerous. It is a moral abdication.”

This is not a fringe position. Dozens of countries have called for a ban on fully autonomous weapons. The UN has debated it for years. But no binding treaty exists. And several major powers—including the United States, China, and Russia—have resisted limits, arguing that autonomous systems could reduce civilian casualties by enabling more precise targeting.

Leo rejects that logic outright. He doesn’t argue that autonomous weapons would be less effective. He argues that they would be wrong regardless of effectiveness. Some decisions cannot be delegated. Period.

This puts the Catholic Church in an interesting position. It’s not pacifist—just war theory exists for a reason. But it’s drawing a hard line at the moment of lethal judgment. A human must pull the metaphorical trigger. Not code.

Christopher Olah’s Uncomfortable Admission

The most surprising moment of the encyclical’s release wasn’t in the text itself. It was on the dais.

Standing next to the Pope—or at least, standing in the same Vatican press room, carefully positioned to show alignment without overstepping—was Christopher Olah, a senior researcher at Anthropic. Olah is not a household name, but in AI circles, he’s respected for his work on mechanistic interpretability (roughly: trying to understand what neural networks are actually doing inside).

Olah’s statement was brief, but it was not typical corporate PR. He said: “Every frontier AI lab operates inside incentives that can conflict with doing the right thing.”

Let that land. A senior researcher from one of those labs, standing next to the Pope, publicly admitting that his own industry’s incentives are misaligned with safety and the common good.

He went on: “We need external accountability. We need voices that are not inside the bubble. The Pope’s moral clarity is a gift, and we should receive it as such.”

This is remarkable for two reasons. First, because Anthropic has positioned itself as the “responsible” lab—more cautious than OpenAI, more transparent than Google DeepMind, more willing to say no to military contracts. For Olah to admit that even Anthropic’s internal incentives are not sufficient is a significant concession. Second, because it signals that Anthropic is seeking external validation and oversight. They are not saying “trust us.” They are saying “hold us accountable.”

Whether that’s genuine or strategic—a way to shape regulation before less friendly regulators do it—is an open question. But it’s striking to hear it in Vatican City, with the Pope as the audience.

Why the Pope, Why Now?

The casual observer might ask: why does the Catholic Church care about AI? Isn’t this a bit outside their lane?

Two answers.

First: scale. The Church has 1.4 billion members. That’s nearly one in six people on Earth. When the Pope speaks on a moral issue, it is heard—not just in pews, but in presidential palaces, corporate boardrooms, and newsrooms. An encyclical on AI ensures that AI ethics is now on the agenda of every Catholic university, every national bishops’ conference, every Catholic NGO. That’s a lot of institutional weight.

Second: the Church has been thinking about technology and human dignity for a long time. Before AI was a headline, there was Rerum Novarum (1891) on the condition of labor. There was Laborem Exercens (1981) on the dignity of work. There was Caritas in Veritate (2009) on economics and human development. The Church’s social teaching is a tradition of asking: does this technology serve the human person, or subordinate the human person to efficiency, profit, or power?

Magnifica Humanitas is not a departure. It’s an extension. AI is the newest form of capital. And the Church has been criticizing unaccountable capital for 130 years.

What the Encyclical Doesn’t Say

A fair critique of Magnifica Humanitas is that it’s long on principles and short on mechanisms.

What does “robust legal frameworks” actually mean? Which level of government? What kind of enforcement? How do you regulate a model that’s trained in one country, deployed via API in a second, and used by a customer in a third?

Leo doesn’t answer these questions. Encyclicals don’t do implementation details. That’s for legislatures, regulators, and courts. The Pope’s job is to set the moral direction. The rest is politics.

But there’s a deeper silence. The encyclical does not address the possibility that AI might already be too far gone. Too embedded. Too profitable. What if the monopolistic control Leo decries is not a bug but a feature of the technology itself—that frontier models require such massive capital, data, and talent that only a handful of actors can ever build them? What then?

Leo’s answer, implicit but clear, is that we have to try anyway. The alternative—acceptance, surrender, adaptation—is not morally acceptable. Some lines cannot be moved.

The Response: Praise, Skepticism, and Silence

Reactions to Magnifica Humanitas have been predictable across predictable camps.

Progressive Catholics are praising it as a bold intervention against corporate power. Traditionalist Catholics are… confused. AI is not abortion or euthanasia. It doesn’t fit neatly into existing culture war categories. Some traditionalist outlets have downplayed the encyclical, arguing that the Pope should focus on “real” moral issues. (A revealing response: as if the structure of the global economy and the nature of autonomous weapons are not real.)

In Silicon Valley, the reaction has been muted but not silent. A few tech CEOs issued carefully worded statements expressing “alignment with the Holy Father’s concerns.” One leaked internal memo from a major lab reportedly said: “We need to get ahead of this before the Vatican starts advising regulators.”

The most interesting reaction came from an unexpected quarter: China’s state-run AI governance institute released a short statement saying it “welcomed international dialogue on ethical AI” and noting that China “has long supported multilateral governance of emerging technologies.” This is diplomatic code for: we don’t want the Pope siding with the US on this, so we’ll play nice.

What Happens Next

Encyclicals don’t change the world overnight. But they change the conversation. And conversations, over time, change policy.

Here’s what to watch in the coming months.

Catholic institutions will start engaging. Catholic universities will host conferences. Catholic development organizations will incorporate AI ethics into their work. Catholic lobbyists in Brussels and Washington will have new talking points.

Some governments will listen. Italy, France, Spain, Poland, the Philippines, Brazil, Mexico—all have large Catholic populations and governments that at least nod to papal teaching. In some of these countries, Magnifica Humanitas will be cited in legislative hearings. In a few, it might even shape actual law.

The military AI debate will intensify. The US Department of Defense is already developing autonomous systems. China is too. The Pope’s explicit condemnation of lethal autonomous weapons gives ammunition to activists and sympathetic lawmakers who want restrictions or a ban.

Anthropic will face new scrutiny. By standing with the Pope, Anthropic has invited a different kind of attention. They will be asked: if you agree with the encyclical, what concrete steps are you taking? Are you refusing all military contracts? Are you supporting binding international treaties? Are you opening your models to independent audits? Words are cheap. Actions are not.

The One Thing to Remember

You can agree or disagree with Pope Leo XIV’s theology. You can think the Catholic Church has no business commenting on technology. You can roll your eyes at another moral plea from an institution with its own complicated history.

But here’s what you cannot do: ignore the signal.

When the leader of 1.4 billion people writes a 40-page document warning that AI is becoming a monopolistic, morally unaccountable force that threatens to turn humans into cogs in an efficiency machine—and when a senior researcher from one of the most important AI labs stands next to him and agrees—something has shifted.

The question is not whether the Pope is right. The question is whether anyone in power has the courage to act on what he’s said.

Leo ends Magnifica Humanitas with a plea, not a threat. He writes: “Let us not wait for catastrophe to teach us what we already know. The greatness of humanity is not in our power to compute, but in our capacity to choose. Choose well.”

It’s a beautiful sentence. Whether it’s a prophecy or an epitaph depends on what happens next.