Did the Pope use AI to write AI horror stories?

It is possible that AI was used to write parts of Pope Leo XIV’s most recent encyclical on the effects of AI on humanity. An analysis and Linch Zhang posted on the LessWrong forum found some passages of A Successful Person to be between 40 percent and 100 percent written by AI, according to the popular AI monitor Pangram.

The document also includes common characteristics found in AI-generated texts, such as greater use of the word “truth” — which is attributed to Anthropic’s Claude — than previous encyclicals, Zhang says. Another person ran the document’s text and section via Pangramfinding that 62 percent of its first title was labeled as artificial AI. When Seaside wrote about 2,000 words of the document through Pangram, saying that 46 percent was written by AI.

However, some parts are labeled as written by humans. Zhang says Pangram has declared some regions to be “0% AI”. The first 20 verses of the last four evangelicals, when passed by Pangram, were 100 percent confident that they were written by man. And the transcripts of Pope Leo’s speeches, passed through Pangram, were also read as 100 percent human.

AI intelligence cannot be fooled. Different AI analytics can show different results, and even if there is agreement there is no guarantee that it is correct. But Pangram is generally respected among AI researchers. In March 2025Pangram said he estimates false claims about human-authored tasks such as artificial intelligence “to be about 1 in 10,000.”

These books are long letters published by the Pope, whose purpose was to provide teachings that deal with the moral and social problems of the time, according to The New York Times. The book is the first by a pope, and the most recent by Pope Francis in October 2024. It is also the first to focus on AI and its various influences, with Pope Leo presenting it in a surprising way along with Christopher Olah, co-founder of the Anthropic movement.

The Vatican did not immediately respond to a request for comment.