Secondhand Book Sales Surge—AI May Be the Engine Behind the Boom
For years the shelves of used‑book stores quietly echoed the classic trade of curiosity, thrift and local romance. Today, a new tide is pushing these trading corners toward a global, invisible marketplace: artificial‑intelligence training. Independent sellers from Northumberland to Edinburgh have started receiving bulk orders that ship thousands of titles to unknown warehouses in a single week—orders that outpace the normal sales of an entire week for many of the firms.
This surge might seem odd, but it gains weight when we consider the legal backdrop. In 2025, a US judge concluded that using copyrighted books to train large language models does not amount to copyright infringement. According to the ruling, the data sets are "exceedingly transformative," a decision that lifted a significant, but disputed, hurdle for AI developers. The court documents said that the bulk purchases were funded by a Canadian firm and that many books were subsequently destroyed in the training process – a dramatic twist that alarms the very vendors who now profit from the steep surge.
Inside the unsealed documents, investigators discovered what the methodology called "Project Panama"—a plan to "destructively scan all the books in the world." Destructive scanning removes a book’s spine so the pages can be scanned en masse, after which the paper is recycled. The process eliminates the physical book entirely, raising concerns about loss of rare editions that no longer appear in the public domain.
The lovers of hardback treasure find themselves in the middle of an ethical tug‑of‑war. Some points to the practical reality that not every copy is essential: popular titles like The Da Vinci Code may have millions of copies yet all could still be two‑sheeted. Yet for a nineteenth‑century textbook or a single surviving copy of a 1780 edition, a single deletion translates into a cultural loss. Bookshop owners in Edinburgh and Newcastle have voiced a particular concern: "It is devastating if an item that has outlived its print run is buried in a pile of scrap paper because an AI has asked for it.”
Following the rulings, prominent AI firms insist that book acquisition specifically for training is common practice and that their internal processes never aim to collapse rare or antiquarian texts. Yet the ordinary bookseller, who remains unsure where their best‑selling paperbacks disappear to, continues to be caught in a moral dilemma: profit versus preservation.
Even as the intellectual‑property debate rages on—U.K. law requires rights‑owner permission for training while U.S. law more readily protects transformations—pedagogical circles are calling for greater transparency. The next step could involve stringent royalty agreements that meet the needs of both authors and the dynamic demands of AI technology.
In short, the secondhand books market has become a front‑line battleground for the future of knowledge-sharing. If the so‑called "booming" sales continue to be sourced by AI developers who intend to pulverize paper, then the value of even the most ordinary print edition may skyrocket—at the cost of the very books that keep us readable and human.
