On May 25, 2026, Pope Leo XIV published “Magnifica Humanitas,” a 42,300-word encyclical that addresses the risks of artificial intelligence and calls on governments to slow down AI development. The document warns of misinformation, conflict, and job disruption driven by AI, and raises concerns that the technology “tends to increase above all the power of those who already have economic resources.”
The encyclical was notable not just for its content, but for the company present at its presentation. Pope Leo XIV released the document alongside Christopher Olah, a co-founder of Anthropic, the AI safety company behind the Claude model.
That detail became more significant when AI detection researchers turned their tools on the encyclical itself. Researchers using the Pangram detection tool estimated that approximately 46% of the content showed markers consistent with AI-generated writing. Some opening chapters scored as high as 62% machine-generated probability. The stylistic marker that drew the most attention: the word “genuinely” appeared nine times throughout the document, compared to zero occurrences in comparable previous papal encyclicals. That pattern is associated with Anthropic’s Claude model specifically.
The Vatican has not publicly commented on the AI authorship claims. AI researcher Yoshua Bengio, a pioneer in deep learning, endorsed the broader dialogue the encyclical opens around AI risks. Tech investor David Sacks questioned what governance safeguards existed around the document’s production. Critics argued that unverified AI authorship could undermine the document’s moral authority.
The irony is pointed. A document that warns of the risks of AI and calls for tighter regulation may itself have been partially authored by an AI from the same company whose co-founder stood next to the Pope at the launch. Whether or not the Vatican used Claude, the situation illustrates the transparency gap that exists between how institutions use AI and how they present their outputs.
For sales teams tired of cold leads, slow customer responses, and manual processes, Dapta is the ultimate tool.
Dapta is the leading platform for creating AI sales agents specifically designed to increase inbound lead conversion. Respond to your leads in less than a minute with voice AI and WhatsApp that converts.
If you want your team to sell more while AI handles the complex stuff, you have to try it.
The Magnifica Humanitas case is a high-profile version of a problem that content teams, marketing departments, and legal firms are increasingly navigating: what are the standards for disclosing AI assistance in writing that is presented as institutional or human voice?
Detection tools like Pangram are becoming more precise at identifying model-specific stylistic markers. The “genuinely” flag for Claude is an example of how these tools can now point to a specific AI system rather than just AI in general. As these tools improve, the gap between “was this AI-assisted?” and “which AI was used?” will narrow.
For enterprises publishing thought leadership, legal briefs, regulatory filings, or any content where authorship credibility matters, a proactive disclosure posture reduces risk. The debate over Magnifica Humanitas will likely accelerate calls for a global standard on AI content disclosure, a policy gap that regulators in the EU, US, and Latin America have been circling without yet closing.
