Did the Pope use AI to write about the dangers of AI?

The buried issue here is not the irony but the reliability of the detector itself. Pangram, like all current AI-text classifiers, carries documented false-positive rates that make any single-document verdict nearly meaningless, and the LessWrong post does not appear to have been peer-reviewed or independently replicated.

This is largely disconnected from recent activity in our archive, as we have no prior coverage of the Vatican, Pangram, or AI-detection tooling to anchor against. It belongs to a broader and underreported space: the credibility infrastructure around AI governance discourse, where the tools meant to enforce authenticity are themselves contested. That gap matters because high-profile misattributions, whether correct or not, can delegitimize serious policy documents and give bad-faith actors a template for dismissing inconvenient institutional positions.

Watch whether the Vatican issues a formal statement on the encyclical's drafting process within the next few weeks. If they confirm or deny AI involvement with specifics, that sets a disclosure precedent other institutions will face pressure to match. If they stay silent, the unverified Pangram result will likely calcify into accepted fact regardless of its accuracy.