Ebook supply rose 19-fold in three years while AI-written books squeezed human authors. The study gives copyright law the market-harm evidence it lacked.
Alpha Score of 49 reflects weak overall profile with moderate momentum, weak value, moderate sentiment. Based on 3 of 4 signals – score is capped at 90 until remaining data ingests.
Amazon's ebook marketplace is getting more crowded, and a new study says AI-written titles are doing the crowding. The number of books recording sales rose 19-fold between 2023 and 2026, while reader spending on those books grew ninefold. The squeeze is in the gap: more titles competing, less revenue per title.
The researchers examined sales data for more than 14,000 ebooks on Amazon from 2023 through 2026, flagging titles with substantial amounts of machine-generated text. Books with no detected AI text lost market share over the same stretch, the researchers found.
The hard part of the copyright fight has been proving market harm. The US Copyright Office concluded last year that using copyrighted works to create commercial content that competes with the originals goes 'beyond traditional fair-use boundaries.' Courts weigh four factors in a fair-use analysis, and one asks whether the new work harms the market for the original. The Amazon data addresses that factor directly.
Anthropic's book-scanning operation sits at the center of the dispute. The company bought millions of used books, cut off their spines, scanned the contents to train its models, and destroyed the originals. A federal judge ruled that process fair use, on the theory that digitizing a book you own does not create an extra copy. The open question has always been what happens after training, when models turn what they learned into new content.
The copyright fight is a first test of a larger liability question: who answers for what a model produces. On the security side, the same question is the subject of AI Rogue Agents: Who Bears Liability for Breaches.
The study's authors point out that AI can produce thousands of books at almost no cost, so human authors end up competing against a flood of new content that wouldn't exist without their own work.
AI companies have said their models do not financially harm the creators whose work they trained on. Ed Newton-Rex, CEO of the nonprofit Fairly Trained, wrote on X that the research leaves that argument 'well and truly dead.'
Drafted by a large language model from the source reporting linked above, then screened by automated publishing checks. It is not read by a journalist before publication. Some articles cite our Alpha Score. Verify prices and figures against the original source. Educational coverage, not personalized advice.