AI Detection Tool Flags Parts of Pope Leo's Encyclical as AI-Written
Automated detectors gave mixed results on the pope’s 42,300-word text, with one run labeling about 46% of sampled passages as AI-written.
- On Monday, automated AI detectors flagged portions of Pope Leo XIV's roughly 42,300-word encyclical Magnifica Humanitas as AI-generated, with Pangram finding an aggregate estimate of 46 percent across roughly 2,000 words and individual paragraphs ranging from 40 to 100 percent.
- Released Monday after signing May 15, the encyclical addresses safeguarding human dignity amid AI advancement and was presented at the Vatican alongside Christopher Olah, Anthropic co-founder, who helped frame AI monopolies, employment, and algorithmic regulation.
- Linguistic analysis revealed 'genuinely' appeared 9 times in Magnifica Humanitas versus 0 in similar-length Dilexit Nos, and 127 em-dashes versus 0 in that document—patterns consistent with Claude, whose Constitution uses 'genuinely' 33 times.
- Pangram reported a false positive rate of approximately 1 in 10,000, yet some paragraphs in the encyclical were rated essentially 0% AI by the same detector, illustrating significant variation; the Vatican has not commented on the findings.
- The encyclical's apparent AI involvement signals the technology is entering mainstream, high-stakes institutional communication, raising transparency questions—a contradiction echoed elsewhere as organizations critique AI while adopting it.
15 Articles
15 Articles
Regardless of whether they are Catholic or not, the contribution of the encyclical of Pope Leo XIV Great Humanity is very important because of the call to regulate the power that with the innovations in artificial intelligence are reaching the great global technological consortia, said President Claudia Sheinbaum. It is a power that has even exceeded that which many governments in the world have.
Did Pope use AI to write his anti-AI encyclical? The internet is debating
Pope Leo XIV’s first major encyclical urges governments to slow down artificial intelligence development, warning of misinformation, conflict and job disruption. But the internet is now focused on a different question altogether: was the Vatican’s anti-AI document itself partially written using AI tools like Claude or ChatGPT?
Coverage Details
Bias Distribution
- 67% of the sources lean Left
Factuality
To view factuality data please Upgrade to Premium
