Independent booksellers report unusually large purchases of obscure and outdated titles, prompting concerns that books could be destined for large-scale digitisation
Independent bookshops in several European countries are reporting a curious new phenomenon: enormous orders containing thousands of books that appear to have little connection with one another.
One particularly striking case occurred in Galway, where Kennys Bookshop received an online order for around 5,000 titles. While the retailer is accustomed to dealing with large purchases, the unusual mixture of books immediately attracted attention.
The selection ranged from serious regional and historical works to older practical guides and titles with little obvious current demand. What made the transaction even more unusual was that such a substantial purchase was placed online without any attempt to negotiate a bulk discount.
Kennys has decades of experience selling books internationally. Its online presence dates back to 1994, and the business regularly supplies institutions and organisations purchasing significant quantities of English-language books. Large institutional customers, however, would typically discuss pricing before completing an order of this size.
Similar Orders Reported in Berlin
The Galway case does not appear to be isolated.
Independent and antiquarian booksellers in Berlin have reported receiving similarly puzzling requests involving large quantities of books, including old editions and titles that would normally attract very limited interest.
Some orders reportedly arrive electronically during the night and can be worth tens of thousands of euros.
For independent retailers operating in an increasingly difficult market, such purchases can provide a welcome financial boost. At the same time, the unusual nature of the orders has created concerns about who is buying the books and what happens to them afterwards.
The growing suspicion among some booksellers is that the material could ultimately be used to provide training data for artificial intelligence systems.
AI Training Fuels Booksellers’ Concerns
Those suspicions have been strengthened by previous revelations about how technology companies have obtained printed material for AI development.
In the United States, reports surrounding Anthropic revealed details of a large-scale programme involving the acquisition of millions of physical books. According to information that emerged through legal proceedings, books were purchased, digitised and used as part of efforts to develop artificial intelligence models.
The controversy became part of a wider copyright dispute involving authors and ultimately resulted in a major legal settlement.
Anthropic has previously explained that its Claude models have been developed using several types of information, including publicly accessible internet content, commercially obtained datasets and internally generated material. The company has also said that purchasing books for AI training is an approach used within the wider artificial intelligence industry and has denied buying and destroying rare or antiquarian works.
ISBN Numbers May Help Automate Purchases
European booksellers nevertheless remain concerned that automated systems could be identifying titles that are missing from existing digital collections.
One theory among German retailers is that ISBN numbers make it relatively easy for automated software to catalogue books and determine which titles have yet to be acquired or digitised.
If purchasing systems are automated, they could potentially search the inventories of numerous independent retailers and place orders for thousands of specific titles with minimal human involvement.
Booksellers say the apparently random nature of some purchases supports their suspicions. Rather than concentrating on commercially valuable or popular works, orders can include outdated manuals, specialist publications and books that have spent years attracting little attention.
Copyright Rules Complicate the Situation
The legal situation may also differ depending on where books are purchased, scanned and processed.
Germany’s publishing and bookselling industry has raised concerns about practices involving the scanning and destruction of physical books. Industry representatives argue that digitising copyrighted works without appropriate permission could violate German copyright legislation.
Booksellers have also noticed what they believe may be a change in how large purchases are handled. Instead of individual books being shipped directly across the Atlantic, some suspect that European addresses are being used as collection points before larger shipments are transported elsewhere.
However, the identity of the purchasers and the final destination of many of these orders remain unclear.
Independent Retailers Face an Ethical Question
For booksellers, the situation creates an unusual dilemma.
A single bulk purchase can represent significant revenue at a time when independent bookshops face intense competition from major online retailers. But accepting the orders becomes more complicated if the books are potentially being used to develop technologies that could transform publishing, writing and the wider creative industries.
In Galway, the concern is particularly focused on intellectual property. Kennys Bookshop has indicated that it would not want to participate knowingly in a process where books are acquired primarily so their contents can be copied and used without appropriate permission.
For now, there is no definitive evidence establishing who is responsible for every unusual European bulk order or precisely what happens to the books after they leave individual stores.
But the scale and unusual selection of titles have convinced some booksellers that these purchases are not simply the work of enthusiastic collectors.
As artificial intelligence companies continue searching for enormous quantities of high-quality information to improve their models, Europe’s independent bookshops may have unexpectedly found themselves at the centre of one of the technology industry’s biggest debates: who should be allowed to use published knowledge, and under what conditions?