AI-Generated Books Flood Marketplaces, Squeezing Human Authors' Incomes

The burgeoning presence of artificial intelligence (AI)-generated books on online marketplaces is creating a significant economic challenge for human authors, potentially diminishing their earnings and reshaping the publishing landscape. This issue gained considerable public attention in March when Hachette, a prominent book publisher, withdrew the horror novel “Shy Girl” from its scheduled U.S. release. The withdrawal followed serious allegations that the novel’s author, Mia Ballard, had utilized AI in its creation. While Ballard has publicly denied using AI for the novel, Hachette’s decision to remove the book from its website, Amazon, and U.K. distribution channels underscored a critical juncture for the book industry: how to effectively manage and respond to the influx of AI-authored content.

Recent research is beginning to quantify the scale of this problem. An Originality.AI report, published last month, analyzed over 2,000 books within the religion category listed on Amazon during the preceding twelve months. The findings revealed that a substantial 1,272 titles, or 63% of the analyzed books, were flagged by their AI-detection software as likely having been written by large language models (LLMs). LLMs are advanced AI systems trained on vast datasets of text and code, capable of generating human-like written content. This indicates a significant portion of newly published religious texts may not be the product of human authorship.

Further investigation by a study led by Stony Brook University, a public research university in New York, examined a larger dataset of 14,419 self-published genre-fiction books sold on Amazon between 2023 and 2026. These books were processed through an AI-detection software named Pangram. The study identified that 2,880 of these books, approximately 20%, contained "substantial" AI text, defined as more than 25% of the book’s prose being flagged as AI-generated. Alarmingly, the research also found that 2,168 books had over 50% of their prose identified as AI-generated, and nearly 1,000 novels were found to be more than 90% AI-generated. This particular study, which has not yet undergone peer review, suggests a direct correlation between the increasing volume of AI-generated books and a decline in authorial revenue.

The research indicates a concerning economic trend: over a three-year period, the total number of books with observed sales on Amazon experienced a dramatic increase of 19.2-fold. However, the corresponding quarterly revenue generated from these sales rose by a significantly smaller margin of 8.9-fold. This disparity suggests a marked decrease in revenue per book sold. The researchers attribute this phenomenon to a "crowding-out effect." Essentially, the sheer magnitude of AI-generated books flooding online marketplaces like Amazon makes it statistically less probable for a customer to discover and select a book written by a human author. This saturation leads to fewer sales for individual human-authored titles and, consequently, reduced income for human writers. The implications of this trend extend beyond individual livelihoods, raising fundamental questions about content authenticity, market fairness, and the long-term economic viability of traditional authorship in an increasingly AI-influenced digital age.