Used bookstores across North America are reporting a rise in unusual bulk book orders, prompting concerns that some purchases may be linked to the growing demand for printed books to train artificial intelligence systems. Booksellers say the pattern of purchases has become increasingly suspicious, with buyers seeking large quantities of books across a wide range of genres, often without the typical interest shown by collectors, libraries, or educational institutions.
Several independent bookstore owners said the orders involve hundreds of books at a time and are frequently placed through intermediaries or newly created accounts. In many cases, buyers are unwilling to explain the purpose of the purchases, raising concerns that the books may be digitized by destroying their bindings so the pages can be scanned more quickly—a practice commonly referred to as “book shredding.”
The growing use of printed books for AI training has sparked debate within the publishing industry. While digitizing books is not illegal in every circumstance, authors and publishers have raised concerns over copyright, fair compensation, and the ethical use of their work in developing artificial intelligence models. Industry groups argue that large-scale scanning without proper licensing could undermine the value of copyrighted material.
Booksellers say the trend is also affecting the availability of used books for regular customers. Some stores have begun limiting the number of books that can be purchased in a single transaction or are asking buyers to disclose the intended use of unusually large orders before completing sales.
AI companies have faced increasing scrutiny over the sources of data used to train their models. Many technology firms maintain that they comply with applicable copyright laws and licensing agreements, while rights holders continue to push for greater transparency and stronger legal protections.
Experts say the issue highlights the growing tension between the rapid advancement of artificial intelligence and the need to protect intellectual property. As demand for high-quality training data continues to increase, regulators, publishers, and technology companies are expected to face mounting pressure to establish clearer rules governing the use of printed and copyrighted works in AI development.