Imagine scrolling through Amazon’s spirituality section looking for a book on Wiccan rituals or Hindu philosophy, only to find that the vast majority of what’s on the virtual shelf may never have been touched by a human writer. That’s the unsettling picture painted by a new study into AI-written religious books, which found that machine-generated text has quietly taken over a huge share of religious and spiritual publishing on the platform.
The research, released this week by AI detection firm Originality.ai, sampled more than 2,000 recently published titles across 14 religious and belief categories sold on Amazon. The conclusion: nearly two out of every three books were likely written by artificial intelligence rather than a person.
Summary
Key takeaways
- Originality.ai flagged 1,272 of 2,034 religious books analyzed on Amazon — 63% — as likely AI-written.
- The witchcraft category The highest percentage was recorded at 78%, with Hinduism reaching 76% and Taoism at 74%.
- Researchers flagged 53% of fact-checkable claims in witchcraft books as potentially false.
- Amazon relies on an honour system for authors to self-disclose AI-written work, with no public data on how many actually do.
- Amazon has confirmed it buys physical books through commercial channels to scan their pages for AI training data.
Majority of Religious Books on Amazon Are Likely AI-Written
The numbers behind this study are striking on their own: 63% of the religious books sampled scored high enough on Originality.ai’s detection model to be classified as likely machine-written. That’s not a fringe result buried in a niche corner of the store — it represents the majority outcome across an entire commercial category spanning multiple world religions.
Originality.ai Study and Methodology
Report author Michael Fraiman said his team analyzed book descriptions, author bios, and text samples from each title, running them through a scoring model. Anything that scored 50 or higher was classified as “Likely AI.” Out of 2,034 books reviewed, 1,272 crossed that threshold.
Fraiman was careful to frame the findings as probability rather than proof. “We don’t make claims to certainty about whether books are definitively AI-written,” he told Decrypt. “Our model determines the likelihood that something was AI-written to a degree of certainty.” He added that researchers spot-checked flagged titles manually, saying the team can “tell the difference between likely AI-written content and human-written content” by this point in their work.
AI Authorship Rates Across Religious Categories
Not every faith tradition was affected equally. The witchcraft category — pulled from Amazon’s Wicca, Witchcraft & Paganism listings — topped the list at 78%, covering titles on healing crystals, herbal remedies, and cleansing negative energy. Fraiman suggested the appeal to publishers is obvious: these books can promise solutions for mental health or “detoxing” without needing to meet any scientific standard.
Hinduism followed at 76%, and Taoism came in at 74%. Sikhism, Buddhism, Islam, Catholicism, Protestantism, Orthodox Christianity, and Judaism landed somewhere in between. Mormonism registered 42%, atheism 40%, and Satanism 22% — the lowest of the categories studied. Researchers cautioned that sample sizes varied considerably between categories, which limits how directly the percentages can be compared against one another.
Verification Challenges and False Positives in AI Detection
A high AI-detection score doesn’t automatically mean a book is misleading — but in the witchcraft category, the study found a troubling overlap between the two. Researchers flagged 53% of fact-checkable claims in witchcraft titles as potentially false, raising concerns about readers turning to AI-generated spiritual guidance for health or wellness advice that hasn’t been vetted by anyone.
Limitations of AI Writing Detection Tools
Detection tools like Originality.ai’s model are built to estimate likelihood, not deliver a verdict. That distinction matters, because AI-detection technology as a category has a track record of inconsistency. Fraiman’s own framing — that the model offers “a degree of certainty” rather than definitive proof — reflects a broader industry acknowledgment that no detector currently guarantees accuracy.
Examples of Conflicting AI Detection Outcomes
The unreliability of AI detectors isn’t hypothetical. In October 2024, a test of four separate tools analyzing the U.S. Declaration of Independence produced wildly different verdicts: ZeroGPT rated it 97.93% AI-generated, QuillBot called it entirely human-written, and GPTZero gave it an 89% probability of human authorship — three tools, three contradictory answers on the same 18th-century document.
The stakes got more serious in March, when Colombia’s Supreme Court rejected a legal filing after a detector flagged it as AI-generated. In an ironic twist, when an attorney later ran the court’s own ruling through the same detector, it came back rated 93% AI-generated. That episode underlines why treating detection scores as courtroom-ready evidence remains legally shaky.
Amazon’s Role and Ethical Questions in AI Content and Training
Amazon sits at the center of two separate but related controversies: how it polices AI-written books sold on its marketplace, and how it sources raw text to train its own AI models. Neither practice currently comes with much public transparency.
Amazon’s Honor System for AI Content Disclosure
According to Fraiman, “Amazon is aware of how common AI-written content is on their platform, and uses the honour system for authors to self-identify their work as AI-written.” Originality.ai did not share its findings with Amazon before publishing, and Fraiman noted plainly: “We have no data on how many authors actually disclose that.” Amazon did not immediately respond to Decrypt’s request for comment.
This matters because self-disclosure without verification leaves a wide gap between what’s technically required and what actually happens. If the AI-written share of religious books really is close to two-thirds, a purely honor-based system may be catching only a fraction of it.
Use of Physical Books for AI Training and Ethical Concerns
Separately, Amazon’s appetite for training data has drawn scrutiny over how it handles physical books. An investigation tracked a shipment of rare books to an Amazon facility in Las Vegas — where workers cut bindings off books and scanned the pages into AI training datasets.
Amazon confirmed that it “purchases books through commercial channels to help develop and improve the products and services our customers use,” though the company’s statement stopped short of confirming AI training specifically.
Taken together, the two threads point to the same underlying tension. On one side, AI-generated text is flooding categories like religious publishing with content that hasn’t been fact-checked and, in some cases, contains false claims. On the other, the raw material used to build those same AI systems is coming from physical books being destroyed with limited public accountability. Neither Amazon’s self-disclosure system for sellers nor its book-sourcing pipeline for training data currently offers much visibility into scale — leaving readers, authors, and booksellers to rely largely on outside investigations to understand what’s actually happening behind the storefront.
FAQ
How reliable is the detection of AI-written religious books on Amazon?
Originality.ai uses a scoring system that indicates likelihood rather than certainty. The researchers caution that the results point to probable AI authorship, not definitive proof.
Which religious book category had the highest proportion of likely AI-written books?
The witchcraft category had the highest rate, with 78% of sampled titles flagged as likely AI-written.
What ethical concerns are raised by Amazon’s use of physical books for AI training?
Amazon buys and scans physical books, sometimes rare ones, for AI training and then discards them, raising ethical concerns about the destruction of source material and the lack of explicit author consent.
Does Amazon require authors to disclose if their books are AI-written?
Amazon uses an honor system where authors self-identify AI-written content, but there is no public data on how many actually disclose this.
Article produced with the assistance of artificial intelligence and reviewed by the editorial team.
