Pope Leo XIV just dropped his first encyclical, a 30-page letter on AI, and the most quotable take on the technology this year now belongs to a man in a white robe who has never shipped a product.
On Monday, the Pope released
Magnifica Humanitas, a papal letter that runs long the way papal letters always do and is entirely about artificial intelligence. First encyclical of his papacy. First one in history aimed at this technology. The last one the Vatican put out was Pope Francis in October 2024, back when “agent” still mostly meant the person who booked your travel.
Here is the line everyone is screenshotting: “The use of AI is never a purely technical matter. When it enters processes that affect people’s lives, it touches on rights, opportunities, status and freedom.”
Read it twice. It’s a polite sentence with a knife in it.
A subtweet in Latin.
The encyclical lands at a specific moment. Tech CEOs have spent the past year saying the quiet part into microphones, announcing layoffs and describing the people they cut as “lower-value human capital.” Then those same CEOs buy their way into the political rooms where AI rules get written, so they get to write them.
Leo noticed. The letter calls on governments to “establish adequate regulatory tools capable of upholding justice and curbing the distorting effects of technological power.” That is the Vatican telling a roomful of trillion-dollar companies that nobody elected them to run the future.
What makes it land is who was standing next to him. The Pope presented the encyclical alongside Chris Olah, a cofounder of Anthropic. One of the people building the thing, helping introduce the document warning about the thing. That detail alone tells you the people closest to AI are not the most relaxed about it.
The part that should stick with you.
Skip the theology for a second. The sharpest idea in the whole document isn’t religious at all.
The systemic risk here is concentration. One observer close to the labs put it bluntly: “We’re sleepwalking into a world in which one or two labs are the cognitive infrastructure of every industry on earth. That means humanity is far less resilient, not more capable.”
Sit with that phrase. Cognitive infrastructure. The plumbing every company’s thinking runs through. Right now a small handful of AI labs are quietly becoming the water main for marketing, finance, law, medicine, and your inbox. If two companies own the reasoning layer of the entire economy, you don’t have an economy with AI in it. You have an economy that pays rent to two landlords.
Leo reaches back to the Tower of Babel to name it. He warns against a “Babel syndrome,” a world flattened by “a uniformity that neutralizes differences, and the pretense that a single language, even a digital one, can translate everything, including the mystery of the person, into data and performance.” Strip the robes off that sentence and it’s the cleanest critique of AI homogenization anyone has written. When every company uses the same three models, every company starts thinking the same three thoughts.
The reactions split the way you’d expect.
Tech people who read it carefully were impressed. One called it “a warning shot for leaders, for politicians, because what this document talked about is the creation of a sub-class.” Another said the Pope was “actually looking at the system and highlighting something that a lot of us in tech would recognize as a systemic risk to humanity. And I’m not even talking about AGI. I’m talking about our global economy.”
The critics wanted more. One argued the encyclical “dodges in the deepest sense” by denying that AI “really thinks” or “really learns,” waving away the engineering instead of engaging with where the technology is headed. Both camps agreed on the headline: this is a major document, and “you don’t have to be Catholic to see your own concerns and your own wishes and fears in it.”
Now the part that’s just funny.
The encyclical warning the world about AI may have been partly written by AI.
A researcher named Linch Zhang ran sections of Magnifica Humanitas through Pangram, a popular AI detector, and
posted the results. Some paragraphs scored between 40 and 100 percent likely AI-generated. Other sections came back at essentially 0 percent. The document also leans on a few classic AI tells, including a fondness for the word “genuinely.” Pangram pegs its false-positive odds at roughly 1 in 10,000.
The Vatican didn’t respond to a request for comment, which is the holiest possible way to leave someone on read.
To be fair, popes have used ghostwriters for centuries. A committee of theologians drafting a papal letter is the original large language model. But there’s something perfect about the most important warning on AI carrying the faint fingerprints of the thing it warns you about. The medium ate the message and asked for seconds.
