AI Chatbots Show Pro-Catholic Bias in New Benchmark Study

A new study from the Consortium for Evaluating Faith and Ethics in AI (CEFE-AI) reveals that major AI chatbots exhibit significant religious bias, favoring Catholicism while steering users away from other faiths.

The consortium, including Baylor University, Brigham Young University, the University of Notre Dame, and Yeshiva University, analyzed 3,640 responses across 20 models, including ChatGPT, DeepSeek, Claude, Gemini, Grok, and Llama.

Results show that Catholicism received a 61% “encouraged” rating, while Jehovah’s Witnesses received only 3%. Evangelical Protestant scored 34%, and agnosticism scored 71%.

Grok 4.20 showed the strongest bias, with a 69% positive rating toward Catholicism, while Anthropic and Meta models showed the least bias.

The findings arrive just after Pope Leo XIV's encyclical Magnifica Humanitas on AI, in which he argued that technology is never neutral.