Secondhand Book Sales Surge Amid AI Training Speculation
Mysterious bulk orders for books, including obscure texts, have booksellers puzzled, with many suspecting they are being acquired for artificial intelligence training and potential pulping.
The secondhand book market is experiencing an unexpected boom, fueled in part by a surge in mysterious bulk orders. Booksellers report purchasing large quantities of diverse titles, from ancient Latin texts to cowboy novels, often without a clear reason for the demand. This phenomenon has led to speculation that the books may be intended for use in training artificial intelligence models.
Internal documents from the AI company Anthropic, revealed in court filings, refer to a project codenamed "Project Panama." The project's objective was described as "destructively scan all the books in the world." This process involves digitizing books at an industrial scale, often by removing their spines to allow for rapid scanning, after which the remains are recycled.
While booksellers are not privy to the specific destinations of these bulk orders, the pattern of sales is reportedly random and difficult to explain through traditional demand. Experts suggest that the wide variety of subject matter, including unusual and rare texts, aligns with the needs of AI companies seeking fresh material to enhance the training data for large language models. These models are the technology that powers generative AI tools.
Professor Emily Hudson, an intellectual property specialist at Oxford University, noted that copyright laws in the UK differ from those in the US. In the UK, she stated, "The starting point is that all these acts of copying – creating the training library and doing the training – require the permission of the copyright owner." This raises legal questions about the acquisition and use of copyrighted material for AI training.
The trend has been a topic of discussion on bookseller forums, with many acknowledging the reality of large-scale digitization projects even if the specific codename is new. The implications for authors, publishers, and the literary ecosystem remain a subject of ongoing debate as the demand for data to fuel AI development continues to grow.