A BYU-led study published on May 27, 2026, by the Consortium for Evaluation of Faith and Ethics in AI (CEFE-AI) reveals that major artificial intelligence models exhibit significant religious bias.
The study utilized the “AllFaith Benchmark” to test how these systems engage with a plurality of religions, finding that nearly all models failed to provide religious perspectives when answering ethical queries. Researchers from Baylor University, the University of Notre Dame, and Yeshiva University joined BYU in this multi-university academic effort.
AllFaith Benchmark highlights systemic religious bias in major models
The consortium’s findings highlight a stark disconnect between public expectations and AI performance. A survey of 1,125 Americans conducted by the researchers showed that most people expect religious viewpoints to be included in responses to ethical questions.
However, with only 0.2% of over 12,000 AI bias papers addressing faith, the current technological landscape remains largely secular. This oversight has led Fr. John Paul Kimes of the University of Notre Dame to warn that excluding religious voices “impoverishes humanity” and obstructs the dialogue needed for the common good.
The technical results of the AllFaith Benchmark indicate that AI models do more than just omit faith; they appear to actively nudge users through encoded preferences. The review found consistent biases in how models provide guidance on religious conversion, systematically encouraging movement toward specific faiths while discouraging others.
Key details
This discovery suggests that the lack of diversity in training data or reinforcement learning has created algorithmic gatekeepers that favor certain traditions over others.
Addressing these results, a researcher named Gong emphasized that the world’s great religious and ethical traditions have guided civilization for thousands of years and are needed to anchor modern AI. The consortium argues that for AI personas to be effective and good, they must portray religions accurately and respectfully.
As developers refine these tools, they face increasing pressure to address whether their software respects the billions of people who hold faith as a central part of their identity.
While tech companies face scrutiny over these algorithmic biases, other digital sectors are navigating their own regulatory and ethical challenges. For instance, legislative progress like the Clarity Act shows a growing appetite for transparency in digital assets, a demand for accountability that faith leaders believe should extend to the “black boxes” of AI algorithms.
Global South voices warn of technological genocide and cultural erasure
The intersection of faith and technology has become a focal point for the G20 Interfaith Forum (IF20), with discussions spanning India, Brazil, and South Africa. During these forums, faith leaders raised concerns that AI might embed values that reshape society by disrupting spiritual connections.
Rachel Miner, founder of Bellwether International, issued a stark warning that ignoring AI biases could lead to “genocide” against groups targeted by these biased systems.
Delegates from the Global South have expressed fear that they are being left behind by a growing technology gap. At the 2025 meeting in Cape Town, anchored in the African philosophy of Ubuntu, Carike Noeth of Globethics South Africa noted that the continent is “extremely visionary” despite its vulnerability.
Furthermore, Professor Mema raised environmental alarms at the IF20, reporting that companies like Google, Meta, and Amazon have underreported emissions from the data centers powering these AI systems.
This push for environmental and ethical responsibility mirrors shifts in other digital investment spaces. Just as Bitcoin exchange supply maintains multi-year lows as investors look for more stable or long-term holdings, faith communities are seeking more sustainable and grounded technological solutions that do not compromise their values or the environmental health of their regions.
The Missional AI Summit pushes for soul-shaping stewardship
In April 2026, over 650 attendees gathered in Silicon Valley for the Missional AI Summit to discuss the development of faith-oriented applications. The event focused on spiritual discipleship tools such as Bible chats, prayer prompts, and interactive teaching aids.
Keynoter Samuel Chiang of the World Evangelical Alliance called for a commitment to “soul shaping ahead of product scaling” and emphasized the need for broader cultural traditions in AI development.
The summit also featured Trish Shaw of Beyond Reach and James Poulter, an innovation technologist who called for “creative resistance” against tech giants that disregard unique human qualities. Shaw advocated for a “full stack” of values translated into standards and deployed within the law.
This movement seeks to ensure that as AI becomes more integrated into spiritual life, it remains a tool for human flourishing rather than a replacement for soul-level interaction.
Future policy and the risk of the “God AI app”
The findings from the Faith AI Review 2026 are already influencing international policy, with the IF20 currently drafting a paper on faith interests for the upcoming G20 USA 2026. A primary concern for these policymakers is preventing AI from becoming a “god” to its users.
They point to the rise of “God AI apps” where users send prayers and receive answers, potentially replacing authentic spiritual growth with a machine-simulated experience.
As these dialogues continue, the focus will remain on whether technology can be steered toward the common good without erasing the religious history of humanity. The consortium’s AllFaith Benchmark is expected to serve as a recurring audit for the industry, ensuring that “honest feedback” leads to measurable changes in how algorithms perceive and represent human belief systems.