Pope Leo XIV's AI encyclical may be partly AI-written

Pope Leo XIV's landmark encyclical *Magnifica Humanitas*, the first papal letter focused entirely on the moral and social impacts of artificial intelligence, may have been partially written by AI itself. A detailed analysis by Linch Zhang, posted on the forum LessWrong, ran the document through Pangram — a respected AI detector with a claimed false positive rate of 1 in 10,000. The results showed certain paragraphs scored between 40% and 100% probability of being AI-generated. Specifically, the first chapter was flagged at 62%, and when The Verge tested a 2,000-word sample, Pangram estimated 46% was AI-written. The analysis noted linguistic markers like an elevated use of the word "genuinely," a known trait in outputs from Anthropic's Claude models.

However, AI detection remains imperfect. Other sections of *Magnifica Humanitas* registered as "essentially 0% AI," and all 20 introductory paragraphs of the last four encyclicals — written by Pope Francis — scored 100% human confidence. A transcript of Pope Leo's speech also passed as fully human. This encyclical is particularly notable not only for its subject matter but because it was presented alongside Christopher Olah, co-founder of Anthropic, raising questions about the Church's relationship with AI developers. The Vatican did not respond to requests for comment. The irony of using AI to write about AI's dangers underscores the complexity of detecting and managing AI-generated text in high-stakes communications.