Recent analyses suggest that parts of Pope Leo XIV's latest encyclical on AI might have been penned by an artificial intelligence. An investigation by Linch Zhang revealed certain sections to be up to 100% AI-generated, according to the AI detector Pangram.

The document contains unusual writing patterns typical of AI outputs, such as a spike in the use of the word 'genuinely,' similar to Anthropic's Claude. A section-by-section check showed that 62% of its first chapter was flagged as AI-generated. When The Verge ran about 2,000 words through Pangram, it estimated 46% were AI-written.

However, the analyses aren't definitive. Other parts of the document are confidently identified as human. The first 20 paragraphs of the last four encyclicals were rated at 100% confidence as being written by humans. A transcript of Pope Leo’s speech was also rated as 100% human.

AI detection remains imperfect, with false positives and varying results across different detectors. Pangram is generally respected among AI researchers, but its accuracy isn't guaranteed. The encyclicals are significant letters addressing moral and social challenges, making the use of AI a complex issue.