The Pope has been warning the world about the dangers of artificial intelligence. According to Pangram Labs' AI detection tool, at least some of those warnings were written by AI.
Pangram Labs updated its Chrome extension - a browser plugin that puts warning labels on AI-generated text as you scroll through social media feeds - and it flagged content on the Vatican's accounts as AI-generated. The irony is hard to miss.
What Pangram's Tool Actually Does
The extension scans text in your social feed and adds a visual indicator when it detects AI-generated content. The detection works by analyzing statistical patterns in text - things like sentence structure, word probability, and stylistic consistency that tend to differ between human writers and large language models (AI systems trained on enormous amounts of text, like ChatGPT or Gemini).
The use case is real. Social media is filling with AI-generated posts and comments that are increasingly difficult to spot. A tool that flags likely AI content directly in your feed gives users more context about what they're reading, which matters when that content is framed as a personal opinion or moral argument.
The False Positive Problem Nobody Has Solved
AI text detectors have a well-documented accuracy problem. Academic integrity services like Turnitin have faced backlash for flagging human-written essays as AI-generated. The core issue: AI detectors look for patterns, and humans who write in formal, structured ways - or who heavily edit their work - can trigger those same patterns.
Vatican communications are written, translated, edited, and approved through institutional processes. That kind of polished, formal output can look a lot like AI text to a statistical model, regardless of who actually wrote it.
None of this means the Vatican's social content definitely wasn't AI-assisted. Many large institutions now use AI to help draft posts and official statements - it would be more surprising if none of them did. The Vatican hasn't commented on the flagged content. But a label from a Chrome extension is not proof of anything; it's a statistical signal that comes with significant uncertainty.
The timing matters. As AI tools become standard at organizations of all sizes - nonprofits, governments, religious institutions - the question of disclosure is increasingly relevant. An AI-assisted statement isn't inherently dishonest, but audiences generally expect to know when AI had a hand in writing something, particularly when that something is a moral or ethical argument. The Vatican's public position on AI has been notably cautious; if AI is being used to draft that position, that tension is worth naming.
For everyday social media users, the Pangram extension adds useful context and will catch a lot of obvious AI-generated content. Just don't treat its labels as verdicts.