Did the Pope’s Anti-AI Encyclical Get (Partially) Written by AI? The Pangram Detector Says 46%

The Vatican just dropped its first-ever encyclical on artificial intelligence — and the internet immediately asked the most 2026 question possible: Was this thing written by the very technology it warns against?

Did the Pope’s Anti-AI Encyclical Get (Partially) Written by AI? The Pangram Detector Says 46% On May 15, 2026, Pope Leo XIV released Magnifica Humanitas (“Magnificent Humanity”), a major social encyclical that frames AI as a serious threat to human dignity, truth, work, and moral responsibility. The document calls for ethical guardrails, warns against autonomous weapons and deepfakes, and insists that technology must remain subordinate to the human person.

Then came the plot twist.

Independent analysis using the respected AI-detection tool Pangram suggests that roughly 46% of the roughly 2,000 words sampled from the encyclical shows strong signs of AI generation — most likely Anthropic’s Claude model. When run paragraph-by-paragraph, the results are uneven: some sections read as fully human, while others are packed with the tell-tale stylistic fingerprints of large language models (repetitive phrasing, overly polished transitions, and an unusually high frequency of certain words like “genuinely”).

The findings first surfaced on LessWrong and quickly spread. The Verge independently tested a substantial excerpt and arrived at the same 46% AI estimate.

The Irony Is Immaculate

Did the Pope’s Anti-AI Encyclical Get (Partially) Written by AI? The Pangram Detector Says 46% What makes the story delicious is the guest list at the official presentation.

The only representative from the tech industry invited to speak at the Vatican event was **Christopher Olah**, co-founder of Anthropic — the very company whose model appears to have helped draft parts of the document condemning the unchecked rise of AI.

You truly cannot make this up.

Not the Whole Thing — But Enough to Raise Eyebrows

Importantly, the Pope’s own spoken remarks during the presentation pass Pangram’s tests cleanly. Those words carry the unmistakable human rhythm and pastoral tone expected from a pontiff. The AI influence seems concentrated in the written text itself — the kind of lengthy, carefully crafted theological document that traditionally demands weeks of human drafting, editing, and prayerful revision.

Did the Pope’s Anti-AI Encyclical Get (Partially) Written by AI? The Pangram Detector Says 46% AI-detection tools are not infallible. Pangram is widely regarded as one of the most accurate available, but even its creators acknowledge false positives are possible, especially with polished, edited text.

No one is claiming the Vatican secretly handed the entire encyclical to ChatGPT and hit “generate.” What the data suggests is something more subtle and human: papal ghostwriters (or their assistants) leaned on AI for drafting, polishing, or expanding certain sections — a practice that has become increasingly common across industries, from journalism to law to academia.

A Very 2026 Scandal

The Catholic Church has spent centuries perfecting the art of moral pronouncements on new technologies. Now it finds itself in the awkward position of issuing a high-stakes warning about AI… while apparently using AI to help write the warning.

The optics are undeniably funny. As one wag on LessWrong put it: “The robot helped write the papal bull against robots.”

Whether this revelation undermines the encyclical’s authority is a theological question for another day. What it does confirm is that in 2026, even the Vatican — one of the oldest and most tradition-bound institutions on Earth — is not immune to the quiet infiltration of generative AI into serious writing.

The machines have already breached the walls. The question now is whether the shepherds noticed… or simply asked Claude for a better first draft.