Pope Leo XIV’s recent encyclical Magnifica Humanitas was supposed to be a sweeping statement on AI’s impact on human dignity. Instead, it’s become a case study in irony.

The 60-page document, released last week, warns that AI threatens to diminish authentic human expression and undermine our capacity for genuine thought. It also appears to have been written, at least in part, by an AI.

The evidence

Linch Zhang, a researcher posting on the forum LessWrong, ran portions of the encyclical through Pangram, a popular AI detection tool. The results were striking: certain paragraphs scored between 40% and 100% likely to be AI-generated.

More damning are the stylistic tells. The encyclical uses the word “genuinely” with unusual frequency, a known signature of Anthropic’s Claude model. It’s the kind of adverb humans reach for occasionally but AI uses as conversational filler, like a nervous tic in code.

Other markers align with AI-generated text: a certain smoothness of prose that never quite catches on anything, paragraphs that flow without the friction of actual human revision, and a tendency toward balanced clauses that sound thoughtful but don’t always say much.

The Vatican hasn’t confirmed or denied using AI in drafting the document. A spokesperson told The Verge only that “the Holy Father’s teaching represents his authentic vision,” which is the kind of non-answer that does more to fuel speculation than quash it.

Why this matters

If the Pope used AI to write about AI’s dangers, that’s not necessarily hypocrisy. Plenty of people use tools they also want regulated. But it does raise a question the encyclical itself never answers: when is it appropriate to automate human expression, and when does doing so undermine the very authenticity you’re trying to protect?

Magnifica Humanitas argues that AI risks “replacing the irreplaceable” by offering synthetic substitutes for human creativity, judgment, and moral reasoning. It’s a reasonable concern. But if that’s true, what does it mean that the argument itself may have been assembled by the machine?

The document includes a memorable reference to The Lord of the Rings, comparing AI developers to Saruman, who “in his pride… believed he could wield power without being corrupted by it.” It’s a pointed jab at the tech industry’s tendency to see itself as neutral toolmakers rather than moral actors.

Except the Tolkien comparison may have been AI’s idea too. The passage flows a little too neatly, and the metaphor is exactly the kind of literary flourish that large language models excel at: recognizable, respectable, and just vague enough to sound profound without requiring a specific argument.

The tech industry responds

Unsurprisingly, the tech world is having fun with this. Peter Thiel, who has spent years publicly wrestling with Tolkien’s legacy and reportedly sees himself as more Gandalf than Saruman, may well be the implicit target of the Pope’s Lord of the Rings reference. Ars Technica published a tongue-in-cheek investigation parsing whether the Gandalf quote was a direct shot at the Palantir co-founder.

If it was, the irony deepens. Thiel’s entire career has been built on the idea that technology isn’t neutral, that it reshapes society in ways its creators can’t fully control. He’d probably agree with much of what Magnifica Humanitas says. He just wouldn’t expect it to come from a document that might have been ghostwritten by Claude.

What happens next

The Vatican could clear this up quickly by releasing a statement about its drafting process. It won’t. Papal communications don’t work that way, and the Church isn’t in the habit of explaining itself to AI researchers on the internet.

But the episode highlights a broader problem: we don’t have good norms yet for disclosing AI use in high-stakes writing. If the Pope can release a major doctrinal statement without saying whether a machine helped write it, what chance do we have of transparency in legal briefs, legislative drafting, or corporate communications?

The encyclical itself calls for “clear attribution” when AI contributes to creative or intellectual work. It’s a sensible standard. The question is whether the people setting it are willing to follow it themselves.