According to a study by Originality.ai, 82% of publications about medicinal plants listed in the herbal medicine section on Amazon are likely written with the help of artificial intelligence. Experts analyzed 558 books published from January to September 2025 and found a massive influx of unmarked and unverified texts on the platform.
This is reported by Business • Media
Fictional Authors and Unreliable Content
Researchers discovered dozens of fictional authors and brands offering dubious recipes and recommendations. In many cases, the books contain advice on preparing remedies such as “ginkgo tincture for memory improvement” or “jellies for boosting immunity,” yet their effectiveness lacks scientific validation. One of the most popular books was the Natural Healing Handbook by Luna Philby, which topped the “Aromatherapy” and “Herbal Medicine” categories. However, journalists could not find any information about the author or brand mentioned in the description, and text analysis indicated a 100% likelihood of it being generated by AI.
“This reveals a vast amount of unmarked, unverified content that has completely flooded Amazon,” said study author Michael Freeman.
Expert Concerns and Calls for Regulation
Experts emphasize that the opacity of the origins of such books poses risks for readers interested in folk medicine. Often, pseudonyms are used in the texts, such as “Rose, Fern, and Carnations,” along with emojis depicting leaves — a typical sign of AI-generated materials. Some books reference dubious and discredited herbalists, including Barbara O’Neill and Alfredo Bowman, known for promoting unverified cancer treatment methods.
“There is now a vast amount of herbal research that is absolute nonsense. AI cannot sift through this junk — it only confuses people,” commented herbalist Sue Sprang.
Den Conway, CEO of the Publishers Association, urged Amazon to implement labeling for books created using AI and to remove unreliable content. Company representatives responded that they adhere to strict standards for published materials and use tools to detect violations regardless of whether the content is created by a human or artificial intelligence.