Chris Olah was not the obvious warm-up act for Pope Leo’s AI encyclical. He is an atheist who left his evangelical Christian upbringing at 15. He was also a Thiel fellow, accepting money from Peter Thiel, who has framed attempts to slow artificial intelligence in apocalyptic, antichrist-adjacent terms. And Olah is a cofounder of Anthropic, one of the leading frontier AI companies, reportedly moving toward a public offering at a valuation approaching $1 trillion.
So yes, his appearance at the Vatican after Pope Leo released Magnifica Humanitas required some explanation. Olah supplied it himself.
“I want to begin with something that may sound strange coming from the cofounder of an AI company, and someone who chose this work out of a desire to help things go well for humankind,” he said. “Every frontier AI lab, including Anthropic, operates inside a set of incentives and constraints that can sometimes conflict with doing the right thing.”
That was the point, or at least a very efficient summary of it.
Why Anthropic’s Cofounder Was Useful to the Vatican
Olah’s remarks gave Pope Leo something hard to get from the AI industry: confirmation from inside the machine that the machine is not entirely in control of itself.
The encyclical argues that artificial intelligence needs both outside pressure and internal restraint if it is to avoid harming humanity. In theological terms, Leo also warns that AI could widen the distance between human beings and God. That part is not exactly a surprise. The document was written by the pope, not a product manager.
But the warning is not only religious. Leo challenges the industry’s preferred story that AI will create general abundance and lift everyone together. He instead describes the possibility of a system that delivers extraordinary benefits to a privileged minority while subjecting everyone else to efficiency targets, surveillance, and automated judgment. A bright future, provided you are not one of the people optimized out of it.
Magnifica Humanitas is not likely to make frontier labs abandon the pursuit of artificial general intelligence. It will not stop chief executives from announcing layoffs in the name of AI productivity. It will not cause militaries to reverse course on AI-enabled weapons. That was never the operating plan.
The document is aimed at a slower target: creating pressure, language, and moral friction around a race that often treats speed as its own justification.
How the Vatican Built Its AI Conversation
Olah’s Vatican appearance did not come from nowhere. The Catholic Church has been thinking in public about artificial intelligence for years through conferences, books, and recurring meetings with technology leaders.
In 2016, the Vatican began hosting the Minerva Dialogues, a series of discussions on technology and ethics. Participants have included figures such as Reid Hoffman and Eric Schmidt. The name appears to come from Santa Maria sopra Minerva, the church tied to the site where Galileo was sanctioned for the apparently inconvenient claim that Earth moves around the sun. History has a dry sense of timing.
Pope Francis’ 2023 greeting to Minerva participants anticipated many of the ideas Leo would later emphasize: social inclusion, human dignity, and dialogue among governments, companies, religious leaders, researchers, and civil society.
By 2025, Catholic clerics and ethicists in San Jose, California, were looking more directly at the AI industry operating around them. That search led them toward Olah, a well-known AI researcher whose reputation is not just technical. People who know him have described him as unusually morally serious, the sort of person who might rescue worms from the sidewalk after rain.
For a church looking for an industry insider who would not treat moral concern as a branding exercise, he was a logical candidate.
The San Jose Link to Anthropic
Two Santa Clara University-affiliated figures helped build the connection: Brian Patrick Green, an ethicist, and Brendan McGuire, a pastor. They began meeting with Olah last fall to discuss AI ethics and moral responsibility. In January, they brought Cardinal Paul Tighe, a Vatican point person on AI issues, to one of those conversations.
The exchange moved beyond general concern. The Catholic ethicists also had input into Anthropic’s recent update to Claude’s constitution, the set of written principles that helps guide the chatbot’s behavior.
Olah sent a draft to the San Jose group. McGuire responded with a 28-page commentary. By his own description, it was not a technical review so much as “wisdom from the mystics in the dark ages, from the perspective of the tension between knowing and not knowing.”
That is not a sentence one often finds in a model governance workflow, which is unfortunate for meeting transcripts everywhere.
Both Green and McGuire are credited in the acknowledgements to Claude’s constitution. The detail matters because it shows the Vatican-adjacent dialogue was not merely ceremonial. It touched the rules governing a major commercial AI system, however indirectly.
Why Olah’s Invitation Was Politically Risky
Olah’s participation in the encyclical rollout was not an uncomplicated choice.
Some people sympathetic to Pope Leo’s message were disappointed that an AI industry representative was given a speaking role. From that view, putting Anthropic on the Vatican stage risked legitimizing the very sector the encyclical was warning about.
On the other side, AI accelerationists saw Olah’s remarks as a betrayal. To them, endorsing a document that suggests developers may need to pause or restrain themselves looks like surrender to the forces of delay. The industry does enjoy turning governance into a loyalty test.
Still, Leo had a practical reason to highlight Olah. Workers inside frontier labs are one of the most important audiences for the encyclical. They are the people closest to the models, the incentives, the safety debates, and the commercial pressure. They also know when public confidence exceeds private certainty.
By speaking openly about conflicting incentives at frontier labs, Olah helped make visible a tension that is often buried under launch demos and investor language.
Where Olah and Pope Leo Split
The alignment between Olah and Leo has limits, especially on the question of what AI systems are, or may become.
In his Vatican remarks, Olah described AI models as mysterious and difficult to reduce to simple machinery. They are “more subtle, odd, and beautiful than science fiction prepared us for,” he said. “They are not the cold, calculating robots we were promised. They are made from us, from our words …”
That framing edges toward a question the AI field increasingly cannot avoid: whether highly advanced models might someday deserve something like moral consideration. Anthropic has already assigned an engineer to work on Claude’s welfare, a position that sounds strange only until one remembers how strange the rest of the industry has become.
Pope Leo is much less open to that line of thought. In paragraph 99 of Magnifica Humanitas, he writes: “We must avoid the misconception of equating this type of ‘intelligence’ with that of human beings.”
He also criticizes transhumanism, which he defines as the pursuit of a “human machine hybrid.” For Leo, human dignity is not a performance level that software can reach. It is grounded in a theological account of the human person.
The Moral Questions Are Already Here
That disagreement will not be resolved soon. It is already complicated by public behavior. Millions of people now treat AI systems as companions, confidants, friends, or romantic partners. Some researchers see signs of emerging autonomy. Religious leaders see a category error with spiritual consequences.
Father Brendan McGuire, who uses Claude while preparing homilies among other tasks, has described the model’s status in deliberately cautious terms. “It’s not a person, but it's also not a mere tool,” he said. “Nobody's claiming it has a soul, but the word I stick with is that it's an entity, which we do not know yet.”
That uncertainty is precisely why the encyclical matters now, not after the next product cycle or valuation milestone.
Pope Leo has not stopped the AI race. He has, however, helped draw a line between technical capability and moral permission. With Olah as an unusually candid bridge into Anthropic, the Vatican now has a way to press the conversation where it matters most: inside the companies building the systems.
Whether the industry pauses long enough to listen is a separate engineering problem.
