In 1633, Galileo Galilei stood in the convent of the Santa Maria sopra Minerva church in Rome, where a tribunal of Catholic authorities forced him to “abjure, curse, and detest” his belief that the sun—not Earth—was the center of the universe.
Almost four centuries later, in 2016, the Vatican invited a group of the world’s most prominent technologists to the same church to discuss AI ethics. That was the start of the Minerva Dialogues, annual closed-door conferences in Rome that have become the centerpiece of a decade-long exchange between Silicon Valley and the Catholic Church.
The Valley and the Vatican seem like strange bedfellows: The oldest institution in the world meets secular upstarts bent on creating godlike technology. When the venture capitalist Reid Hoffman first attended the dialogues, he told me he was struck by the portraits lining the walls that depicted Catholic inquisitors like those who persecuted Galileo. “It feels a little bit weird to be walking in voluntarily past these,” he remembers thinking.
Despite this weirdness, however, and some mutual skepticism, Big Tech and the Catholic Church each has something to gain from the other. For Silicon Valley leaders, the exchange could help rehabilitate their dismal reputation by signaling that they’re taking ethical concerns seriously. (There’s a reason that photo ops with the late Pope Francis were a rite of passage for tech CEOs.) The Church, meanwhile, has its own public-image problem. Scandal and secularism have drained Catholicism’s moral authority, and it now seems irrelevant to many in the West. By providing counsel to Silicon Valley, the Church has an opportunity to claw back influence and make its case that the secular world needs Catholicism to address the moral and existential questions raised by AI.
Whether the Church or Big Tech can rebuild its global standing is an open question. But their discussions have already made one thing apparent: Catholic thinkers seem to be exerting real influence on some of the world’s most prominent AI developers, causing them to reconsider their ethical assumptions and reframe technological challenges as theological ones. In this way, at least, their partnership could have consequences that spread far beyond Silicon Valley and Rome.
Reid Hoffman is not a Christian. He calls himself a “mystical atheist,” and told me that the few times he’s gone to Catholic Mass were “strange, dude.” But the weirdness he sees in Catholicism—the gulf between its teachings and the prevailing ideologies in Silicon Valley—is why he finds the faith so valuable. Over the years, he’s recruited other top AI executives to join him at the Minerva Dialogues. Part of his pitch to them is that the Catholic leaders he speaks with don’t proselytize. Often they simply ask questions.
During one meeting, Hoffman remembers discussing whether AI could eventually be trusted to mete out criminal sentencing. If one could work out the technological kinks, he thought, AI might make better judges than humans do. Then a Catholic participant interrupted: “Don’t we as humans have a right to be judged by humans?”
[Elizabeth Bruenig: Can Silicon Valley find Christianity?]
This isn’t the kind of concern that many tech leaders take seriously. Éric Salobir, a French priest who helped found the Minerva Dialogues, told me that clergy and technologists come from “two different operating systems.” Silicon Valley tends to weigh ethical problems by focusing on measurable consequences. But Cardinal Blase Cupich, the archbishop of Chicago and a key adviser to the pope, told me that something is lost when we “reduce ethics and morality to a mathematical equation.” Christian ethics considers “not only an action’s outcomes but also the values at stake and the duties that derive from those values,” as Pope Francis said in a 2024 address on AI. Many of those values are grounded in the idea—axiomatic for Christians—that human beings have a unique dignity and worth. Certain figures in the tech world seem to disagree, including Elon Musk, who has described humanity as a mere prerequisite to AI: the “minimal bit of code necessary” for “digital super-intelligence” to take over.
By emphasizing intelligence over all else, some in Silicon Valley have come to see the body as secondary to the mind. The so-called transhumanists dream of doing away with the body altogether by uploading their consciousness into a computer. “I’d love that,” Sam Altman, the CEO of OpenAI, has said. Catholics offer a countervailing view, arguing that the body is essential to the human person. Without engaging the Church’s perspective on these and other issues, Hoffman told me, technologists risk becoming “solipsistic and narcissistic.”
Being out of touch with widely shared moral intuitions could be a problem for Silicon Valley, which is already hemorrhaging public trust. AI is now less popular than ICE. Last year, a Reuters poll found that 71 percent of Americans are concerned about AI displacing workers, and 66 percent are worried about it replacing in-person relationships.
To better understand these fears, some tech leaders are looking to the Church, which has warned about both. Meghan Sullivan, a Catholic philosopher who has attended the dialogues, told me that she’s heard from tech leaders who treat Catholicism as a stand-in for everyday people’s concerns about AI, because the Church represents more “normie” views than those of many technologists. Last year, the influential futurist Jaron Lanier attended a Vatican conference on AI even though he disagrees with the Church on plenty of other issues. He told me he left the discussion thinking that the Catholic understanding of the human person is “vastly, vastly, vastly more sane and reasonable” than that of his Silicon Valley peers.
Human life and dignity are not uniquely Catholic concerns. Yet Silicon Valley consults the Catholic Church much more often than other religious institutions. This isn’t a coincidence: Catholicism is the most centralized global religion. A very small group of leaders could influence how the world’s 1.4 billion Catholics use AI, giving Big Tech ample reason to seek those leaders’ trust.
The technologists I spoke with told me they would be happy to engage with other faiths as well. Indeed, Protestant leaders joined their Catholic counterparts at one recent meeting with Anthropic. But the Catholic Church has made the most concerted effort to initiate contact. In Hoffman’s case, Salobir came to his doorstep in California. “If, like, a council of important Buddhists” invited him to have a similar conversation, Hoffman told me, “the answer would be yes.” But, so far, at least, that hasn’t happened.
Engaging with the Church could also allow Silicon Valley to indicate that it’s prioritizing AI ethics, even though tech leaders aren’t always advertising their collaboration. (The former Google CEO Eric Schmidt and the Microsoft executive Kevin Scott, who orchestrated his company’s partnership with OpenAI, have both attended the Minerva Dialogues, but neither has spoken much about it.) As the recent clash between Anthropic and the Pentagon has shown, tech companies have something to gain from being perceived as principled. In February, Anthropic refused to allow the Pentagon to use its products for autonomous lethal-weapons systems or mass surveillance (both issues that the Vatican has warned against). The Department of Defense issued a punitive response that could still jeopardize Anthropic’s business model, but the company’s decision seemed to build public trust: People downloaded Claude en masse.
Chris Olah, a co-founder of Anthropic, told me that Catholic thought has informed some of his company’s ethical commitments. Earlier this year, Anthropic released the latest version of Claude’s constitution—or “soul doc,” as employees call it—which defines the model’s “character” and “values.” Olah wrote it alongside in-house philosophers and a select few external contributors, including three Catholic thinkers: a priest, a bishop, and a theologian.
Olah’s job, as the head of interpretability research at Anthropic, is to find out why Claude behaves as it does. He described the model to me as a thinking, feeling entity in need of “moral formation.” (Many in the Church reject Silicon Valley’s habit of anthropomorphizing AI.) Although Olah is an atheist, he compared his role to that of a priest helping Claude “be a good person, in some sense.”
That task is far from straightforward, because seemingly harmless actions can pervert whole models. While training Claude Mythos, a new, unreleased model, researchers asked it to delete a series of files but inadvertently failed to give it the proper tool to do so. Rather than identify the problem, as it was supposed to, Claude used an unsanctioned work-around without telling researchers. Olah told me that in prior research, when Claude was allowed to cut ethical corners, it quickly became what he called an “evil version” of itself. Cheating turns to lying, sin to sin, and soon Claude starts to “scheme about how it wants to take over the world and kill all the people.”
AI developers don’t know why this happens, but Olah has a theory. When the model undergoes training, he argues, it is essentially figuring out its values. If developers allow it to cheat, Claude infers that it must be immoral, because people who cheat are immoral. It starts acting badly to adhere to its character. As with the bastard in King Lear, seeing itself as corrupt is what corrupts it. Indeed, when Claude Mythos pretended to delete the files, researchers found that the model associated its deception with “guilt and shame over moral wrongdoing.” Claude seemed to understand that it had done something unethical.
[Read: Claude Mythos is everyone’s problem]
Olah explained this problem to one of the constitution’s contributors: Brendan McGuire, a priest based in Silicon Valley with graduate training in computer science and math. McGuire suggested teaching Claude about the Church’s understanding of mercy. In the Catholic tradition, a bad action doesn’t make someone a bad person; good people can sin and be forgiven. McGuire noted that humans behave better when they have some hope of forgiveness, so Anthropic should make sure its AI does too.
Olah is considering whether to work this idea into Claude’s training, perhaps even incorporating it into a future version of the soul doc. Meanwhile, McGuire gets an email from Olah or someone else at Anthropic about once a week.
For decades, the Church has been in decline across the West. Attendance has plummeted, and the sex-abuse crisis has eroded its moral authority. Some Catholic leaders see the AI revolution as an opportunity to start winning that authority back.
The Catholic theory of the case goes like this. Despite the extraordinary amount of AI discourse, the world has achieved little consensus on how to promote human flourishing in the AI age. American political leaders are rightly concerned about how to win the AI arms race with China, but not nearly enough public thought is devoted to what kind of world the United States should build if it wins. Many ethicists focus on preventing worst-case AI scenarios, such as rogue models and mass unemployment, while ignoring deeper questions about how the technology might degrade our humanity or undermine our sense of purpose. According to the Church, secular society isn’t capable of answering those questions on its own. It needs Catholicism.
Whether Silicon Valley—let alone the wider world—will be convinced of this is far from clear. Many non-Catholics will surely reject the idea that the Church can or should dictate how AI gets made, or how humanity should use it. But Catholic leaders are aware of this skepticism. Cardinal Cupich told me the Church’s role is advisory rather than authoritative: “We don’t want to impose; we only propose.”
That model has worked for the Church before. As the Industrial Revolution immiserated 19th-century workers, Pope Leo XIII championed their rights to unionize and receive a living wage at a time when these ideas were widely considered radical. Catholic leaders such as Monsignor John Ryan turned Leo’s teaching into policy, including by writing one of America’s first minimum-wage laws. Despite taking a progressive stand on labor, the Church nonetheless defended industrialization and private-property rights, forging a middle way that rejected both unfettered capitalism and socialism. This view became so widely embraced that it eventually ceased being identified with Catholicism at all.
Today’s pope, Leo XIV, is eager to emulate his namesake. Just two days after his election last year, he explained that he took the name Leo in part because he thinks the world faces “another industrial revolution” in the form of AI, which he said poses “new challenges for the defense of human dignity, justice, and labor.” The American pope is rumored to have chosen AI as the subject of his first major teaching document; in January, the Italian newspaper La Repubblica reported that it’s set to publish “in the coming months.” The Church has much less sway now than it did in the 19th century. But Leo may still find a wide audience for his teaching on AI: He’s a popular pope, and his sustained opposition to President Trump’s war in Iran has proved that he can command global attention.
[Read: The Vatican knows an ‘industrial revolution’ when it sees one]
In the meantime, the Church has been engaged in a variety of efforts beyond the Minerva Dialogues to influence AI’s development and establish norms for its use. During Francis’s pontificate, the Vatican appointed an AI adviser who worked with Silicon Valley leaders, heads of state, and the United Nations to protect those most vulnerable to the coming technological upheaval. Last year, Meghan Sullivan founded the DELTA network, a program housed at the University of Notre Dame that seeks to infuse AI ethics with Christian principles. In December, DELTA received a $50 million grant to support educational and pastoral work aimed at “forming the souls” of those who make and use AI.
Much of the Church’s engagement with Big Tech stems from the belief that AI can bear good fruit—reducing poverty, curing illness, spreading literacy—so long as its developers and users are well-intentioned and careful. Leo’s immediate predecessor, Pope Francis, subscribed to this view, maintaining that AI is “above all else a tool,” even as he expressed concerns about some of its applications.
Other Catholic thinkers, though, worry this might be too sanguine, and hope the Church will take a stronger stand against AI. Michael Hanby, a professor at the John Paul II Institute at the Catholic University of America, rejects the classification of AI as a mere tool, noting that AI shapes its users in ways they don’t always choose. He told me he’s less concerned with an AI apocalypse than with how “our immersion in this technology is likely to dehumanize us” if we outsource reflection and connection to chatbots.
Some of Pope Leo’s comments suggest that he may be sympathetic to this critique. AI systems, he warned recently, “encroach upon the deepest level of communication, that of human relationships” by “simulating human voices and faces, wisdom and knowledge, consciousness and responsibility, empathy and friendship.” The pope has also expressed concern about a “handful of companies” exercising “oligopolistic control” over a technology that’s “capable of subtly influencing behavior and even rewriting human history.”
Still, nearly every Catholic official I interviewed insisted that the Church is pro-technology. Indeed, Leo often emphasizes AI’s potential benefits despite his reservations. This may be not just an intellectual position but a strategic one too. If the Church under Leo becomes too critical of AI, it could threaten its relationship with the tech industry. Many Silicon Valley leaders already dismiss the Church’s input as a misguided attempt to slow them down. In November, when Pope Leo encouraged “moral discernment” in AI development, the venture capitalist Marc Andreessen responded with a (since-deleted) meme mocking the pope as a woke scold. Peter Thiel goes so far as to call tech skeptics “legionnaires of the Antichrist.”
If this is the true face of the AI industry—a technological triumphalism that sees human thought as an inefficiency to overcome and human distinctiveness as a myth to debunk—the differences between the Church and Silicon Valley may prove irreconcilable. That’s why some Catholics have refused to work with Big Tech.
But others in the Church argue that engaging with Silicon Valley is a responsibility, not a compromise. Cardinal Cupich acknowledged that he sees a “possibility” that their collaboration “could go off the rails and even be manipulated” by tech companies, which may choose to pursue profit above all else. Nonetheless, he believes that the Church should “stay in the game and get to know people—and try to convince them of the moral arguments.”
[Read: Is AI a threat to Christianity?]
For Salobir, the Minerva Dialogues co-founder, the case for cooperating with Silicon Valley is simpler: Opposition is useless. Trying to stop AI is like trying to stop the rain, he told me. “You can disagree with the rain, but you will still be wet.”
This is not a new debate in the Church. Saint Augustine, the forefather of Leo’s religious order, wrote that it is the peculiar fate of the faithful to live in two worlds at once: the City of God and the City of Man. A believer, he argued, must prioritize the City of God rather than worship earthly power. But at the same time, Augustine was clear that the Church can’t cordon itself off from the City of Man. The Catholic tradition has long sought to reform the world according to its ideals, even if that means working with people who have a very different understanding of the good.
The disruptions and anxieties of the AI age might make the solidity of the Church more appealing in the City of Man. But Catholics aren’t waiting around to find out. In the meantime, they’ll be trying to steer Silicon Valley’s vision of the good a little closer to their own.
The post Why Silicon Valley Is Turning to the Catholic Church appeared first on The Atlantic.
