Did the Pope use AI to write his encyclical?

๐กLearn how AI detection tools are being used to audit the authenticity of high-profile institutional documents.
โก 30-Second TL;DR
What Changed
Linch Zhang analyzed the encyclical using the Pangram AI detection tool.
Why It Matters
This highlights the increasing difficulty in distinguishing human-written text from AI-generated content in formal communications. It raises questions about the authenticity of institutional documents in the age of LLMs.
What To Do Next
Use multi-model detection tools like Pangram or GPTZero to verify the provenance of text in sensitive or high-stakes documents.
Key Points
- โขLinch Zhang analyzed the encyclical using the Pangram AI detection tool.
- โขThe document shows a high frequency of the word 'genuinely', a trait associated with Anthropic's Claude.
- โขPangram analysis suggests 40% to 100% AI-generated content in specific paragraphs.
- โขThe findings highlight the ongoing challenge of verifying authorship in high-profile documents.
๐ง Deep Insight
Web-grounded analysis with 20 cited sources.
๐ Enhanced Key Takeaways
- โขPope Leo XIV's first encyclical, "Magnifica Humanitas," signed on May 15, 2026, and released on May 25, 2026, specifically addresses the critical issue of safeguarding human dignity in the era of artificial intelligence.
- โขThe Pope's personal presentation of the encyclical, a departure from traditional protocol, included the notable presence of AI experts, such as Anthropic co-founder Chris Olah, signaling a direct engagement between the Vatican and the AI industry.
- โขInitial analyses suggest that the Italian version of "Magnifica Humanitas" may contain the highest proportion of AI-generated content among its translations, leading to speculation that it could have been the original draft.
- โขThe Pangram AI detection tool, utilized in the analysis, boasts high accuracy rates (99.8-100%) and near-zero false positives, as validated by independent studies from the University of Chicago and University of Maryland, making it a robust tool for identifying AI-generated text from models like Claude.
- โขBeyond the frequent use of "genuinely," other linguistic markers characteristic of AI writing, particularly from Anthropic's Claude, include a consistent deployment of em-dashes and the repeated use of tricolons.
๐ Competitor Analysisโธ Show
| Feature/Metric | Pangram AI Detector | GPTZero | Originality.ai |
|---|---|---|---|
| Accuracy (False Positives) | Near-zero false positives, 99.8-100% accuracy (University of Chicago validated) | Mentioned as a more widely known consumer-facing tool | Mentioned as a comparable tool |
| Detection Method | Analyzes structural, stylistic, and semantic patterns using NLP and a massive dataset of human/AI writing; focuses on statistical patterns and regularity | Evaluates reliability and accuracy in distinguishing human from AI-generated academic content | Evaluates reliability and accuracy in distinguishing human from AI-generated academic content |
| Target Users | Teachers, publishers, trust and safety teams, academic screening, compliance checks, editorial gatekeeping | General consumers, educators | Educators, content creators, web publishers |
| Cost/Pricing Model | Five free credits/day; subscription plans start at $15/month for 600 credits | Credit-based pricing model (implied as an alternative to Pangram) | Credit-based pricing model (implied as an alternative to Pangram) |
| Updates/Models Covered | Constantly updated with samples from latest AI models (ChatGPT, Gemini, Grok, Llama, Claude) | Analyzes ChatGPT versions 3.5, 4, and 4o | Analyzes ChatGPT versions 3.5, 4, and 4o |
| Explainability | Provides rationale for flagging, flags phrases/terms and likelihood | Not explicitly detailed in search results | Not explicitly detailed in search results |
๐ ๏ธ Technical Deep Dive
- Pangram's AI detection tool employs natural language processing (NLP) and is trained on an extensive dataset comprising both human and AI-generated writing samples.
- It identifies AI-generated text by analyzing structural, stylistic, and semantic patterns, focusing on how closely the writing adheres to statistical norms associated with synthetic language.
- The tool places significant emphasis on detecting structural regularity, consistent sentence construction, repetition of syntactic forms, pacing across paragraphs, and overall consistency in phrasing.
- Pangram provides granular feedback, flagging specific phrases or terms that are highly indicative of AI generation and quantifying the likelihood of AI involvement.
- To maintain its efficacy, Pangram continuously updates its database with writing samples from the latest and most capable AI models, including ChatGPT, Gemini, Grok, Llama, and Claude.
- Anthropic's Claude Opus 4.7, implicated in the encyclical's analysis, was developed through pretraining on diverse datasets (public internet, private, and synthetic data) followed by substantial post-training to align its behavior with ethical guidelines.
๐ฎ Future ImplicationsAI analysis grounded in cited sources
โณ Timeline
๐ Sources (20)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
Weekly AI Recap
Read this week's curated digest of top AI events โ
๐Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: The Verge โ

