63% of Religious Books on Amazon Are Likely AI-Written, Study Finds – Decrypt

Nearly two-thirds of religious books sampled on Amazon were likely written by artificial intelligence, according to a new study of more than 2,000 recently published titles.

In a report released Wednesday, AI detection firm Originality.ai flagged 1,272 of 2,034 books, or 63%, as likely AI-written across 14 religious and belief categories.

Witchcraft had the highest rate, with 78% of books sampled from Amazon’s Wicca, Witchcraft & Paganism category classified as likely AI-written. The titles covered subjects including healing crystals, herbal remedies, and cleansing “negative energies.” According to report author Michael Fraiman, many of the witchcraft books focused on those subjects.

“One can picture the ideal customer as someone looking for solutions to heal themselves, improve their mental health, or ‘detox’ from commonplace chemicals and drugs,” Fraiman told Decrypt. “These Wiccan books can address these concerns without being held to any scientific standard.”

Hinduism ranked second at 76%, followed by Taoism at 74%, with Sikhism, Buddhism, Islam, Catholicism, Protestantism, Orthodox Christianity, and Judaism falling in between. Mormonism came in at 42%, atheism at 40%, and Satanism at 22%. Researchers cautioned that sample sizes varied significantly between categories.

Fraiman said researchers analyzed book descriptions, author bios, and samples, classifying text that scored 50 or higher as “Likely AI.” He cautioned, however, that the results indicate likelihood, not proof.

“We don’t make claims to certainty about whether books are definitively AI-written,” he said. “Our model determines the likelihood that something was AI-written to a degree of certainty.”

Fraiman noted that researchers also spot-checked books flagged as likely AI-generated.

“By this point, our research team can tell the difference between likely AI-written content and human-written content,” he added.

While Fraiman expressed confidence in the findings, AI detectors have produced false positives and conflicting results.

In October 2024, a test of four AI detectors produced conflicting results when analyzing the U.S. Declaration of Independence. ZeroGPT found the text to be 97.93% AI-generated, while QuillBot found it entirely human-written and GPTZero gave it an 89% probability of being human-written.

More recently, in March, Colombia’s Supreme Court rejected a legal filing after detectors flagged it as AI-generated. When an attorney submitted the court’s ruling to the same detector, it was rated 93% AI-generated.

Amazon did not immediately respond to a request for comment by Decrypt.

“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,” he said, adding that the company did not share its findings with Amazon. “We have no data on how many authors actually disclose that.”

The use of physical books to train AI has also drawn scrutiny as developers buy millions of titles, strip them apart, scan their pages into training datasets, and discard the originals.

In August, an investigation tracked a shipment of rare books to an Amazon facility in Las Vegas, where workers cut off their bindings and scanned the pages for AI training data. Amazon confirmed that it purchases books through commercial channels for the operation.