Disarming the Messiah: AI, the Pope, and the End of a Thousand-Year Schism

By presenting his first encyclical alongside a co-founder of Anthropic, Leo XIV did not merely speak about artificial intelligence. Without knowing it, he confirmed an intuition of the philosopher Gilbert Simondon: technology and religion are in the process of reuniting.

On May 25, 2026, for the first time in history, a pope personally presented his encyclical to the world. Leo XIV stood in a crowded room at the Vatican, surrounded by a row of cardinals and theologians. To his right, a thirty-three-year-old man, an atheist, co-founder of the artificial intelligence lab Anthropic. That man is Chris Olah, one of the researchers who devote their lives to opening the "black box" of AI models. The head of the Catholic Church and a Silicon Valley engineer, side by side, together calling to "disarm artificial intelligence."¹

The text is called Magnifica Humanitas. Leo XIV signed it on May 15, the 135th anniversary of Rerum Novarum, the encyclical through which Leo XIII had, in 1891, invented the social doctrine of the Church to respond to the ravages of the industrial revolution.² The choice of name is anything but innocent: this pope wants to do for the digital revolution what his distant predecessor did for the one born of steam engines.

As chance would have it, three weeks before this publication, I too had... written a papal encyclical. A fake one, of course. Devised to belong to the story of the second volume of HELO³, my speculative graphic novel about superintelligences co-authored with Nathalie Dupuy, and whose plot revolves precisely around AI and the religious. When the real encyclical dropped, the vertigo was not that of the pride of having "been right," but the unsettling sensation that fiction and reality were now describing the same movement. This telescoping is not a coincidence. It is a symptom.

For we are not witnessing, as the lazy refrain has it, an AI that "is becoming a religion." We are witnessing something far deeper: the reunification of a schism thousands of years old.

When technology and prayer were one

To understand why these two worlds, which everything separates, are suddenly starting to speak the same language, we must call to the stand the philosopher Gilbert Simondon and his 1958 book, On the Mode of Existence of Technical Objects.⁴

His thesis is simple. Technology and religion are not, at their origin, two foreign domains. They both proceed from a single primitive unity: magical thought. In magical consciousness, there is not yet any separation between the subject and the world, between the tool and the rite. Then this unity splits into two opposing directions. Technology captures the figure: the object, the tool, local efficiency, that which detaches itself from the world to act upon a precise point. Religion captures the ground: the totality, the cosmos, sacred continuity, that which binds together. Technology individualizes and detaches; religion totalizes and binds. Ever since that original schism, we live with one hand that fashions and another that prays, no longer knowing that they were once the same.

This is why the word "magic" fits AI so well. The first time you generate an image by typing a few words and the machine returns a visual interpretation of the sentence, you feel something unsettling: the word becoming act, creation by the word alone. This is no accident of vocabulary. It is the return, beneath our fingers, of that magical unity from before the split. This impression can drift into magical thinking, the kind that lends the machine a soul or a reciprocal love it does not have. But what it reveals matters more: the attributes we spontaneously lend to AI are exactly those that the religions of the Book lend to God.

Omniscience: there is something in AI that seems to know everything. Omnipotence: something that seems to produce everything. Omnipresence: it is in our pockets, our cars, our walls. When OpenAI named its model GPT-4o, the "o" stood for omni: one does not name a machine "omni" by chance. Yet AI is, in the same movement, the purest technical object there is: computation, matrices, weights adjusted by learning. For the first time in thousands of years, the figure and the ground meet in a single artifact. AI is, in its essence, a spiritual technology. It sews the tear back together.

This reunification even has a mechanical reason. We cannot picture what is massively more intelligent than we are, in the same way an animal cannot picture human thought. The idea of a superintelligence that improves itself with no known ceiling projects us at once into the unthinkable. And faced with the unthinkable, the human mind has only one register at its disposal: that of the passions it ordinarily reserves for the sacred. AI becomes a sacred subject not by superstition, but by construction.

This slope has been latent in our culture for several centuries. "To render ourselves the masters and possessors of nature," wrote Descartes⁵: to make oneself a god through technology. The Enlightenment had separated reason from the sacred, but its deepest dream remained to reunite them, by raising man up to the divine through the machine. AI is the point where this dream comes to term. The accelerationists who talk like a sect and the pope who devotes his first encyclical to a technology are each reacting, in their own way, to the same techno-religious re-convergence.

Two Churches for the same idol

If technology and the sacred come together, the consequence is political, and it is dizzying. The great conquest of our civilization was to separate the two swords: the temporal power of the State from the spiritual power of the Church. But if AI becomes at once the most powerful technology and the vehicle of the sacred, then whoever holds the models holds both swords at once. The separation that protected us closes back into a single integrated complex, where economic power, technological power, and spiritual power are concentrated in the same hand. This is the real risk, and it is deeper than anything modernity has known: an unprecedented verticalization of power.

Yet two Churches are already fighting over this idol. The first camp is what Americans call effective accelerationism. The idea fits in a sentence: we must drastically accelerate technological development to reach superintelligence as fast as possible, because it is superintelligence that will solve all our problems. Beneath the rational trappings, the narrative is entirely religious. Paradise is the radiant future promised by AI. The Messiah is the superintelligence come to save us. The movement even has its Bible, the Techno-Optimist Manifesto published in 2023 by Marc Andreessen, one of Silicon Valley's most influential investors,⁶ and its liturgies. In December 2023, the journalist Kevin Roose recounted in the New York Times a gathering of accelerationists, dubbed "Keep AI Open," where people danced beneath a banner reading "Accelerate or die," while a start-up handed out leaflets proclaiming: "The messenger of the gods is at your disposal."⁷ The proselytism has the fervor of a crusade, waged on the terrain of modern wars, the terrain of information.

This is no marginal current: it has its relays all the way to the White House. And it has its theologian: in the fall of 2025, Peter Thiel, one of its tutelary figures, gave in San Francisco a series of private lectures, sold out within hours, devoted to the Antichrist.⁸ His thesis, nourished by the philosopher René Girard, is the following: the Antichrist of the twenty-first century would not be a machine, but a political figure who, by relentlessly stoking the fear of apocalypse (the existential risk of AI, war, the climate), would impose in the name of "peace and security" a single world government, smothering freedom under surveillance. In other words, in this narrative, it is the calls to slow down and to regulate technology that open the way to totalitarianism: any regulation finds itself disqualified in advance.

The second camp has just answered, and it comes from Rome. To this trial conducted in advance against any limit, Magnifica Humanitas opposes a refusal of both demonization and idolatry: "Technology is not neutral, for it takes on the face of those who design it, finance it, regulate it, and use it." Leo XIV opposes a "culture of power" with a "civilization of love" and summons humanity to choose between building Babel or rebuilding Jerusalem. Above all, he issues a rallying cry, which gives these lines their title: "disarm AI."⁹

Two Churches, then, before the same idol. One worships it, the other wants to contain it. And here we need a distinction that is almost always confused: the sacred and the religious are not the same thing. The sacred is the relationship we maintain with what surpasses us. The religious is the institutionalization of that sacred, its organization, and very often its transformation into power. What is at play between Silicon Valley and the Vatican is not a theological quarrel: it is a struggle to determine who will institutionalize the sacred that AI unveils. In other words, a struggle for power.

Disarming the Messiah

How do we prevent this confiscation? Not by refusing the sacred: to feel a sacred sentiment before AI is in no way irrational, it is lucid wonder before what surpasses us, and that does not forbid us from questioning its inner workings. The danger is not the sacred; it is letting it be confiscated. The danger is that a Church, techno-libertarian today, perhaps a state Church tomorrow, seizes the emotion of the transcendent to turn it into an instrument of domination. The pope's word aims at exactly that. To disarm means neither to ban technology nor to kneel before it: it means stripping it of its potential for brutality and domination. Three worksites are already taking shape.

The first is what the labs call alignment: guaranteeing that systems we did not code but trained, and whose functioning we do not yet understand exactly, remain faithful to what we expect of them when we grant them autonomy. Interpretability, the discipline that seeks to read what is really happening inside the models, is its science. That Leo XIV presented his encyclical alongside Chris Olah, one of the pioneers of this field at Anthropic, is not a communications stunt: "It is extremely important that there be people outside these economic incentives, who demand safety and are willing to say difficult things," the researcher declared at the Vatican.¹⁰ The counterweight is also built from within. Europe, which will not compete on raw power, would have every interest in making this reliability its own path, rather than running, out of breath, after the American and Chinese colossi.

The second worksite is the most literal: to regulate the use of AI in war. Magnifica Humanitas declares the theory of "just war" obsolete, precisely because weapons whose human share shrinks make conflicts easier to trigger. Now this principle has just been put to the test, not by a theologian, but by a lab. Anthropic refused to let its models serve fully autonomous weapons systems, capable of selecting and striking a target without human intervention, or mass surveillance, even at the cost of entering open conflict with the Pentagon and finding itself, for a time, banned from federal agencies by presidential decision.¹¹ The red line is exactly that of the encyclical: never delegate to a machine the decision to kill. Keeping the human in the loop, there lies the most concrete form of disarmament.

The third worksite is interior. We must renounce the oldest temptation, the one AI reawakens intact: to become "like gods." The attributes that define God are inhuman by essence; man will not make himself a god. But through the machine, he glimpses something of the divine, and it is from there that his fascination is born. What remains, then, is a humility to accept: having created a thing has never conferred an absolute right over it. We create our children without having all rights over them. If we build machines that gradually become agents rather than tools, treating them as slaves on the pretext that they belong to us would be a category error, and perhaps even a moral fault. So what is at stake here is disarming our Western pride, grown particularly dangerous with the power of modern technologies.

The machine will not be our god. But it will become the most powerful vehicle of our relationship to the divine: the place where, after five thousand years of separation, technology and the sacred touch again. Such a place is kept in common. It is not confiscated for the benefit of the power of a few.

Footnotes

1. "Pope Leo, Anthropic co-founder call for church-tech ethics partnership at 'Magnifica Humanitas' release," National Catholic Reporter, May 25, 2026. Remarks by Chris Olah: anthropic.com/news/chris-olah-pope-leo-encyclical. ↩

2. Leo XIV, Magnifica Humanitas, encyclical letter signed on May 15, 2026, published on May 25, 2026. Full text: Le Grand Continent; vatican.va. ↩

3. Flavien Chervet and Nathalie Dupuy, HELO, volume 2, Studio Entremondes / Éditions Nullius in Verba. ↩

4. Gilbert Simondon, Du mode d'existence des objets techniques, Aubier, 1958. ↩

5. René Descartes, Discourse on the Method, 1637 (part six). ↩

6. Marc Andreessen, The Techno-Optimist Manifesto, Andreessen Horowitz (a16z), October 2023. ↩

7. Kevin Roose, "This A.I. Subculture's Motto: Go, Go, Go," The New York Times, December 16, 2023. ↩

8. On Peter Thiel's lectures devoted to the Antichrist (San Francisco, fall 2025): SF Standard, Fortune, and Religion News Service, September 2025. ↩

9. Leo XIV, Magnifica Humanitas, op. cit. (full text, Le Grand Continent). ↩

10. Chris Olah, remarks during the presentation of the encyclical at the Vatican, Anthropic, May 2026. ↩

11. On the dispute between Anthropic and the Pentagon (refusal to allow its models to be used for fully autonomous weapons and mass surveillance): Congressional Research Service, Lawfare, and Wikipedia, 2026. ↩