The Church and Claude

(The Vatican is taking a lead in addressing one of the most critical issues of our time. This is not just a theological act, it is also a geopolitical move.)

Pope Leo XIV signing the Magnifica Humanitas in May 2026 where he addresses the challenge of AI. Photo source: Catholic World Report.)

The photograph circulated fast. On one side of a lectern in the Vatican’s Synod Hall, Pope Leo XIV in white cassock. On the other, Christopher Olah — thirty-three years old, atheist, co-founder of an AI company called Anthropic — in what appeared to be an off-the-rack suit he had not quite broken in. Between them, a shared microphone. The cardinals arranged behind them looked, depending on your sympathies, either curious or bewildered.

This was May 25, 2026. The occasion was the public launch of Magnifica Humanitas, the first encyclical in papal history devoted entirely to artificial intelligence. What made it doubly historic — and sent the meme factories into overdrive — was that Leo had chosen to present it himself, another first, and to do so alongside an avowed non-believer whose company makes the AI system under discussion. It was, you could fairly say, a lot to explain.

The cynics got there first, of course. PR stunt. Silicon Valley buying indulgences. The Vatican doing what it always does when power shifts: sidling up to whoever holds the new levers. These readings are not entirely wrong. They are just not entirely interesting, either, because they miss what is genuinely strange and significant about the moment — namely, that the relationship between Rome and this particular corner of San Francisco runs considerably deeper than a press conference.

Start with the name. Anthropic. The adjective form of anthropos, the Greek for human. In a boardroom full of companies called things like DeepMind and OpenAI — names that gesture toward the cosmic or the abstract — Anthropic planted its flag in the human. This is not coincidence. The company was founded by people who left OpenAI specifically because they were worried that the race to build more powerful AI was outpacing the race to understand it, and that “understanding” here meant something closer to moral philosophy than to engineering. Their flagship model is named Claude. Their governing ethical document is called the Claude Constitution. The people they consulted while writing it included, among others, Bishop Paul Tighe — the Vatican’s secretary for culture and education — and two Silicon Valley priests affiliated with a Jesuit university.

So when Leo XIV stood next to Olah in the Synod Hall, the visible oddness of the scene concealed a less visible truth: that the ethical framework already running inside Claude had been partly shaped by people who report, ultimately, to Rome.

This is the kind of detail that gets lost in the larger noise about religion and technology being fundamentally incompatible, which is mostly a story that lazy thinkers tell each other. The actual history suggests something more complicated, and more interesting.

A Very Old Habit

Consider the Stamperia Vaticana, the Vatican’s own printing press, founded in 1587 by Sixtus V. Gutenberg’s invention was, by this point, barely a century old. The Catholic Church — which would spend the next several centuries being accused of technophobia — was in fact among the earliest institutional adopters of the most consequential communications technology of the previous hundred years. It understood, with the pragmatic intelligence of an institution that had already survived fifteen centuries, that whoever controlled the printing press shaped what educated Europeans thought. And so it built one.

Jump forward to 1931. Vatican Radio goes on air, its inauguration overseen personally by Guglielmo Marconi — the inventor of radio, not merely some engineer he’d hired. Pius XI had recruited the man who made the medium possible to make it work for Rome. Within a few years, the station was broadcasting in dozens of languages to every continent, and it preserved its editorial independence through Fascism and the Second World War. Today it reaches people in 47 languages, staffed by more than two hundred journalists in sixty-one countries.

Notice the pattern: not hostility to new technology, but rapid, serious, institutional adoption of it. Television? Pius XII allowed a documentary about his life to be made in the 1950s, just as the medium was beginning its conquest of Western living rooms. John Paul II published what is generally considered the first document in Christian history on the cultural implications of computer technology. When Vatican.va went live in 1995, it was one of the earliest institutional websites anywhere. Francis — whose papacy has been shadowed by his own legendarily limited relationship with smartphones — was considerably sharper about the structural implications: in 2015, he merged nine separate Vatican media organisations into a single Dicastery for Communication. Vatican Radio, the Television Center, the Press, L’Osservatore Romano, the Photography Service, the Publishing House, the Holy See Press Office, the Pontifical Council for Social Communications, the Vatican Internet Service — consolidated under one command, with a unified editorial logic. A corporate restructuring indistinguishable in its methods from what any competent media CEO would have done.

What the Vatican has always understood is that the instrument and the ideology it carries are different things. The printing press was welcome; Lutheranism was not. Radio was welcome; Nazi propaganda was not. The internet was welcome; the corrosion of authority it enabled was not. You can think this distinction is self-serving — the Church embracing each new medium until it finds the medium harder to control than expected — and there’s something to that. But you can also think it reflects a genuine and sophisticated institutional intelligence about the difference between the tool and the use made of it. The Vatican has been making that distinction for four hundred years. It has now arrived, with the same logic, at AI.

The Groundwork: Antiqua et Nova

Antiqua et Nova, the doctrinal note issued in January 2025 — “Ancient and New,” titled from Matthew’s Gospel — was the groundwork. Its argument, at base, was philosophical: that artificial intelligence is a product of human intelligence, not a form of it. This sounds like semantic fussing until you realise that the entire commercial mythology of AI depends on blurring precisely this line. The industry names its systems after humans, gives them human voices and human faces, asks them to express opinions and emotions, and then wonders, in breathless press releases, whether they might be conscious. The Vatican’s response — that intelligence, properly understood, involves not computation but understanding, not processing but experiencing, not information retrieval but love — is not pious hand-waving. It is a philosophical position with a two-thousand-year pedigree, grounded in Aristotle’s claim that all people by nature desire to know, and in Augustine’s observation that the restless human heart finds no rest except in God.

The document’s most pointed argument was that AI, misunderstood, risks becoming a new form of idolatry. Secular audiences tend to skip past this language, which is exactly why they miss the sharpness underneath it. Idolatry, in the theological tradition, is not simply the worship of statues. It is the attribution to created things of value and power that belong only to the uncreated. The idol is always a substitute — something made to fill the space that is properly occupied by something else. Applied to AI: if we treat systems that process text as entities capable of wisdom, relationship, and conscience, we are not merely making a category error. We are filling the hole that used to be filled by religion with something that has none of religion’s actual content. The empty calories of ersatz transcendence.

Bishop Tighe, who co-authored Antiqua et Nova, is an Irish priest who has spent years at exactly this intersection — the Vatican’s representative in Silicon Valley’s institutional life, the man who helped build the Institute for Technology, Ethics and Culture at Santa Clara University, the figure who apparently sat down with Anthropic’s people and talked through what values ought to be baked into Claude. He is, if you think about the actual wiring of the Vatican–AI relationship, far more important than almost anyone has noticed.

The Encyclical: Babel and Nehemiah

Then came Magnifica Humanitas, and the category shifted entirely.

A doctrinal note is what the Church issues when it wants to say something clear and carefully reasoned. An encyclical is what it issues when it wants to say something with the full weight of the magisterium behind it — when it wants to establish, for the long term, what the Church teaches. The last time a pope used an encyclical to announce that a major civilisational development constituted a new “social question” requiring theological analysis, it was 1891, and the development was industrial capitalism. Leo XIII’s Rerum Novarum (New Things) founded what became Catholic Social Teaching: a 135-year tradition of engagement with labour rights, property, poverty, and democracy.

Leo XIV’s choice to open Magnifica Humanitas by commemorating the 135th anniversary of Rerum Novarum is not an act of modesty. It is a claim. AI gets the same treatment that industrial capitalism got in 1891. It is the new social question. The same framework that generated Catholic opposition to child labour, the same tradition that produced papal advocacy for workers’ rights and living wages, is now being aimed at transformer models and large language systems and the handful of private companies, richer than most governments, that are building the infrastructure of the digital future without democratic mandate.

The encyclical organises itself around two biblical images that Leo works hard and well. The Tower of Babel: humanity at its most technically impressive, building with one language and one purpose toward the heavens, achieving only the collapse of communication and the dispersal of community, because the project excluded God and substituted efficiency for genuine human relationship. And Nehemiah: the exile returning to a ruined Jerusalem, not imposing a grand vision but distributing sections of wall to families, artisans, priests, young people — rebuilding through shared responsibility, with God at the centre, relationships first and stones after.

Neither of these is a rejection of technology. Babel is a rejection of a relationship to technology — built on the fantasy of self-sufficiency, on the pretence that a single language, even a digital one, can translate everything including the mystery of the person into data. Nehemiah is an embrace of the other kind of building: inclusive, accountable, oriented toward those who have least rather than those who have most.

This distinction gives the encyclical its political teeth. Leo is not wringing his hands about chatbots. He is making a structural argument about power: that the “main drivers of development” in AI are “private, often transnational, parties” whose resources “surpass those of many governments,” and that this creates a situation of private power over collective destiny that is, by the Church’s own principles, illegitimate. “New monopolies of AI” is not a phrase that softens on reflection. It is a direct accusation. And the encyclical’s declaration that the traditional just war doctrine — the framework of Augustine and Aquinas, refined over fifteen centuries — is “now outdated” in the face of autonomous weapons systems capable of delegating lethal force to algorithms, is the single most consequential sentence in a document full of consequential sentences.

The Engineer at the Lectern

Into this argument steps Olah, who is, by all accounts, a genuinely unusual person. His academic work on neural network interpretability — the effort to look inside AI systems and understand what they are actually doing, rather than treating them as black boxes that produce outputs — is considered foundational by people who know the field. He is not, in the way of many tech executives, primarily interested in the money. He is primarily interested in the problem: whether it is possible to build AI systems whose values are genuinely legible, and whose operation can be understood and steered by people who are not AI researchers.

At the Synod Hall, he said that AI safety is “an unsolved problem” and called on religious communities to provide “moral voices that the incentives cannot bend.” This is the statement of a man who understands what markets cannot do. Markets are extraordinarily good at optimising for stated preferences. They are constitutionally incapable of asking whether those preferences ought to be held in the first place. The question of what values should be encoded in systems that will increasingly mediate human life — in healthcare, in education, in warfare, in democratic discourse — is not a question that market mechanisms can answer. It requires a tradition of moral reasoning, a claim about human nature that is not reducible to consumer demand, and an institution with the credibility and the reach to make that claim stick.

The Church is looking at Anthropic and thinking: these are people who, unlike most of their competitors, have actually asked the question. The fact that they consulted Bishop Tighe while writing their ethical constitution is not nothing. The fact that Catholic scholars filed an amicus brief defending the company when the Trump administration tried to ban it from federal contracts — in retaliation, Anthropic said, for refusing to allow its systems to be used in autonomous weapons or mass domestic surveillance — suggests that the relationship has moved beyond courtesy.

There are complications. Anthropic has not signed the Rome Call for AI Ethics, the Vatican-supported pledge whose signatories include IBM, Microsoft, and representatives of Christianity, Islam, Judaism, and the Eastern religious traditions. The reason for this omission is unexplained, and it leaves an awkward gap in an otherwise tidy narrative. Neither party is entirely without self-interest here. This is how all significant alliances work.

The Pews

Which brings us to the question of whether anyone is sitting in them.

The data is messy, as data about religion usually is. What is clear is that the narrative of inevitable, total, irreversible secularisation is no longer as solid as it seemed a decade ago. France — “the eldest daughter of the Church,” a country that spent much of the twentieth century leading the European project of aggressive de-Christianisation — recorded 17,800 adult baptisms in 2024, a 45 per cent rise from the previous year, with the sharpest growth among students and young professionals. Austria saw 85 percent more baptisms in 2024 than in 2023. In the United States, Catholic adult initiates numbered 34,501 in 2024, up from 30,000 the year before, with the trend continuing into 2025 and beyond.

The Pew Research Center, which is professionally sceptical of religious revival narratives, contests some of the more dramatic claims — particularly those generated by opt-in polls rather than head counts at church doors. The “Quiet Revival” in Britain, proclaimed loudly by the Bible Society, looks rather less dramatic when you count actual Anglicans rather than survey respondents who identify as churchgoers. But even the cautious reading supports the conclusion that something has shifted in the cultural weather, especially among the young, and especially in the direction of Catholicism and Pentecostalism rather than the declining mainline denominations.

The interesting question is not whether this is happening but why. The generation that came of age with algorithmic recommendation engines, with social media optimised for engagement rather than truth, with systems designed by engineers who understood people primarily as data points — this generation has experienced, viscerally, what it feels like to be processed. It has grown up inside the very Babel the encyclical describes. And some portion of it is looking for something that insists, against the evidence of its entire digital life, that it cannot be reduced to its data. The algorithm cannot love you. The Catholic Church says it knows something that can.

The Rejected Stones

There is a passage near the end of Magnifica Humanitas that keeps returning to mind. Leo XIV writes that “the rejected stones — the poor, the sick, the migrants and the least among us — will become the cornerstone.” The image is from Psalm 85, and the encyclical echoes the Magnificat — the hymn that gives it its title — in which Mary sings of the mighty cast down and the humble raised up. This is not, in context, merely pious sentiment. It is a criterion of judgment. The measure of any technological future is what it does with the people who are least able to benefit from it and most exposed to its harms. Not the shareholders. Not the early adopters. The ones who get the job losses without the productivity gains, the surveillance without the convenience, the environmental costs without the electric cars.

By this criterion, the current trajectory of AI development fails. The encyclical says so plainly. And the Church — an institution present in the slums of Manila and the refugee camps of Lebanon and the rural poverty of sub-Saharan Africa, as well as in its marble halls — has a standing to say so that the ethics teams of technology companies do not.

Whether any of this matters depends on whether the people building these systems are willing to be constrained by something other than their own judgment and their investors’ expectations. Olah’s presence at the Synod Hall suggests at least one of them might be. His argument that the labs need “informed critics who will tell them when they are failing” and “moral voices that the incentives cannot bend” is either the most sophisticated public relations exercise in recent tech history, or it is a genuine reckoning with the limits of what technical intelligence, on its own, can achieve.

The Vatican, having spent five centuries turning every new communications technology into an instrument of its continuing mission, is betting on the latter. It built a printing press in 1587. It put Marconi on the radio in 1931. It launched a website before most corporations knew what a website was. It merged nine media organs into one in 2015. And on May 25, 2026, it put its pope and a Silicon Valley co-founder on the same stage to announce that the moral framework of the AI age is, in some part, Rome’s business.

The algorithm has brought the Almighty back into the room. Rome noticed first, and it did not come empty-handed.