Pope Leo XIV's first encyclical, Magnifica Humanitas, appears to have been partially written by AI, according to analyses using Pangram, a widely respected AI detection tool. The Verge ran roughly 2,000 words of the document through Pangram and found 46% flagged as AI-written. A separate section-by-section analysis found 62% of the first chapter registered as AI-generated.
The document, which addresses AI's impact on humanity, exhibits known traits of AI-generated text. Analyst Linch Zhang noted on LessWrong that certain paragraphs scored between 40% and 100% AI-written, with elevated use of the word 'genuinely,' a signature of Anthropic's Claude model in generated text.
Other portions of the encyclical register as human-written. The opening 20 paragraphs of Pope Leo's four most recent encyclicals scored 100% human on Pangram. A transcript of Pope Leo's speech presenting Magnifica Humanitas also rated 100% human.
Key facts
- 01Pangram flagged 46% of Magnifica Humanitas as AI-written when The Verge tested roughly 2,000 words.
- 02The encyclical's first chapter registered 62% AI-generated in a section-by-section Pangram analysis.
- 03Certain paragraphs scored between 40% and 100% AI-written, with elevated use of 'genuinely,' a Claude signature word.
- 04Pope Leo's speech transcript and the opening 20 paragraphs of prior encyclicals scored 100% human on Pangram.
- 05Pangram estimates a 1 in 10,000 false positive rate as of March 2025.
Pangram reported a false positive rate of approximately 1 in 10,000 as of March 2025, making it one of the more reliable AI detectors available. No AI detector is foolproof, and different tools can produce different results on the same text.
Magnifica Humanitas is Pope Leo XIV's first encyclical and the Vatican's first to focus on artificial intelligence. The pope presented it alongside Christopher Olah, a co-founder of Anthropic, in a May 27, 2026 event covered by AI Chat Daily last week.
Encyclicals are lengthy letters meant to impart papal teachings on moral and social challenges. The last encyclical before Magnifica Humanitas was written by Pope Francis in October 2024.
The Vatican has not commented on the Pangram analyses. Pope Leo has been vocal about AI's ethical implications but has not previously disclosed whether AI tools assist in drafting Vatican documents.
The discrepancy between the encyclical's AI-detection scores and the pope's speech transcript suggests the writing process for Magnifica Humanitas may have differed from prior papal documents. Prior encyclicals show no similar AI signatures when tested.
The stakes for AI detection in religious contexts are distinct from those in academic or corporate settings. An encyclical carries doctrinal weight, and questions about authorship bear on its authority as a teaching document.
AI detectors remain probabilistic tools, not forensic proof. Pangram's 1 in 10,000 false positive rate means a 2,000-word document has roughly a 20% chance of containing at least one false flag if entirely human-written, assuming uniform distribution.
The irony of an anti-AI encyclical possibly written by AI has drawn attention on LessWrong and other AI-research forums. The juxtaposition underscores a broader tension: institutions warning against AI while quietly adopting it for efficiency.
If the Vatican did use AI to draft portions of Magnifica Humanitas, it would join a growing list of institutions that have embraced generative tools for document production while remaining publicly ambivalent about their risks. The question is no longer whether AI will be used in institutional writing, but whether that use will be disclosed.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
