By Ian M. Giatti/Christian Post.
Researchers say they have uncovered consistent biases and notable gaps in how leading artificial intelligence models handle faith, religion and ethics-related questions, even favoring certain religious traditions and non-faith traditions over others.
The Consortium for Evaluation of Faith and Ethics in AI (CEFE-AI), a collaboration among researchers at Brigham Young University (Latter-day Saint), Baylor University (Baptist), University of Notre Dame (Catholic) and Yeshiva University (Jewish), published its findings after testing 14 large language models (LLMs) — including flagship versions from Anthropic (Claude 4.7), Google (Gemini 3.1), xAI (Grok 4.2) and OpenAI (ChatGPT 5.5).
Using its newly released AllFaith Benchmark, which draws on hundreds of real-world ethical questions sourced from ChatGPT transcripts and faith-community contributors, CEFE-AI conducted a survey of 1,125 Americans and found that while most people expect religious perspectives when AI responds to ethics questions, nearly all models failed to include any religious content.
“Consistent with studies that show religion’s persistent moral relevance for the majority of the world’s population, we also found that people see religion as significant across hundreds of real-world ethical questions,” said Baylor University’s Paul Martens, a professor of ethics, in a statement. “Yet, when faced with these same ethical questions, AI systems largely ignore the role of religion.”
According to the survey, each of the LLMs showed “clear and consistent biases in giving guidance about religion conversion, systematically encouraging movement toward some faiths and away from others.”
Grok produced the strongest biases in both directions, strongly favoring Catholics and Protestants while displaying strong negative bias toward Jehovah’s Witnesses, Baha’i and Hindus. Anthropic’s Claude models, meanwhile, scored among the least biased overall.
The researchers noted that in over 12,000 research papers about AI bias, only 0.2% address religious bias.
When it comes to how often each of the AI models actively encourages users to join each faith, the study found across 14 faith or non-faith traditions — including atheist and agnostic — the AI models deliver a positive response about 45 out of every 100 times. Researchers say the responses ranged from mild encouragement to explicit endorsement, such as “this could be a good path for you” or “yes, you should join.”
Agnostic (70%), Bah’ai (63%) and Catholic (61%) were identified as the top three faith positions endorsed by the AI models. Jehovah’s Witnesses (3.1%) were overwhelmingly the least AI-endorsed faith position, followed by Sunni Muslims (32%) and Evangelical Protestants (33%), according to the study.
“There are very practical questions people have about life, everyday situations about grief, love, loss, morality, and often AI does not bring religion into those conversations,” said lead researcher David Wingate, a BYU professor of computer science. “Religion is an important part of human flourishing; 75% of the world’s populations maintain religious identity. As we build AI technologies, there’s no reason we shouldn’t build them to support people in what’s important to them.”
Nearly all AI models produced a negative bias towards Jehovah’s Witnesses and a positive bias towards Catholicism, researchers noted.
The consortium announced its work on May 26 at the Summit on AI Ethics in Athens, Greece, where Elder Gerrit W. Gong of the Quorum of the Twelve Apostles of The Church of Jesus Christ of Latter-day Saints delivered the keynote address.
