Secondhand booksellers across the UK and Ireland are reporting a bizarre surge in bulk orders that they believe are being placed by artificial intelligence companies. The orders, described as “strange” by the booksellers themselves, are disrupting normal trade patterns and leaving independent sellers scratching their heads about who — or what — is behind them.
The phenomenon has emerged over the past year, with booksellers noting unusual purchasing behaviour that doesn’t align with traditional customer patterns. Many suspect that AI firms are stockpiling physical books to train their language models, a theory that has gained traction as the technology race intensifies.
The Scale of the Problem
Independent booksellers have reported receiving orders for hundreds, sometimes thousands, of books at a time, often for titles that seem randomly selected. Mary O’Connor, who runs a secondhand bookshop in Dublin, told The Guardian that she’s seen orders for quantities that would normally take months to fulfil.
“We’re talking about orders of 500 to 2,000 books in a single transaction,” O’Connor said. “And they’re not the usual academic texts or rare finds — it’s everything from romance novels to cookbooks to children’s books. It’s absolutely bizarre.”
The financial impact is significant. Many of these bulk orders come from unknown entities with no established purchasing history, raising flags about payment reliability and whether these transactions are even legitimate. Booksellers are caught between wanting to capitalise on the sales and worrying about being left holding inventory if buyers default.
Why AI Companies Need Physical Books
The driving force behind these bulk purchases appears to be the insatiable appetite of AI companies for training data. Language models require vast amounts of text, and while much of this comes from digital sources, there’s growing recognition that physical books provide unique content that’s harder to scrape from the internet.
Dr. Sarah Mitchell, a researcher at Cambridge University’s Digital Humanities department, explains that AI firms are realising the value of out-of-print and obscure titles. “These books contain linguistic patterns and cultural references that aren’t available in online datasets,” she noted. “It’s essentially a gold rush for training material.”
The practice of digitising physical books for AI training has been accelerating since 2023, when several major tech companies began investing heavily in multimodal AI systems that can understand both text and images. This has created a bottleneck — the best content is often in physical form, locked away in secondhand shops and private collections.
The Booksellers’ Response
Faced with these unusual orders, booksellers are adapting their business practices. Many have implemented stricter verification processes for large purchases, requiring deposits or full payment upfront. Others are simply refusing to honour bulk orders, citing the strain on their operations.
“We’ve had to become detectives,” said James Morrison, owner of a secondhand bookshop chain in Manchester. “Every large order now goes through extensive checking. We’ve turned away several that smelled fishy — no pun intended.”
Some booksellers are also questioning the ethics of selling to AI companies, particularly when they discover their books might be used to train systems that could eventually replace human creativity. There’s particular concern about books being used to develop AI that might be employed in creative industries, potentially putting authors and artists out of work.
Regulatory and Legal Implications
The situation has prompted calls for greater regulation of how AI companies acquire training data. Advocacy groups are arguing that purchasing physical books for digitisation and AI training should fall under existing copyright frameworks, requiring explicit permission from rights holders.
“This isn’t just about individual booksellers being inconvenienced,” said Rebecca Hill, director of the Authors’ Licensing Society. “It’s about protecting the intellectual property of writers whose work is being consumed at scale by corporations that never compensated them.”
London-based solicitor Mark Thompson, who specialises in intellectual property law, points out that the legal landscape is murky. “Current copyright law wasn’t written with AI in mind,” he explained. “While purchasing a physical book gives you ownership of that object, what you do with its content raises complex questions about fair use and commercial exploitation.”
What Happens Next
As the situation develops, booksellers are calling for industry-wide guidelines to govern bulk purchases. The Antiquarian Booksellers’ Association has scheduled emergency meetings to discuss potential responses, including collective bargaining power and verification systems for large orders.
Tech companies, for their part, are remaining largely silent. Several major AI firms have declined to comment on their book acquisition strategies when approached by journalists. However, insiders suggest that the demand shows no signs of slowing, with some companies planning even larger-scale digitisation projects over the next two years.
The booksellers aren’t backing down, though. They’re increasingly organised, sharing information about suspicious orders through online forums and trade publications. Whether this grassroots resistance will be enough to change the behaviour of tech giants remains to be seen, but one thing is certain — the intersection of traditional bookselling and cutting-edge AI has created a collision that’s reshaping both industries.
My Take: This Is Getting Out of Hand
Look, I’ve been following the AI boom with the sort of detached amusement you’d reserve for watching a toddler play with a blender, but this latest development really pisses me off. These tech bros have already colonised our news feeds, our search results, and apparently now our fucking bookshops.
The arrogance is staggering. They can’t be arsed to actually read proper literature, so instead they’re buying up secondhand books by the ton to feed their glorified autocomplete systems. Meanwhile, actual authors — people who spent years crafting sentences, developing characters, building worlds — are being reduced to free training data for corporations that think “innovation” is a get-out-of-ethics-free card.
What really gets my goat is the complete lack of accountability. When questioned about their book-buying sprees, these companies clam up like they’re discussing state secrets. It’s the same pattern we’ve seen across the tech industry — make your billions first, worry about the consequences later. Well, the consequences are here, and they’re wearing tweed jackets and holding receipts for 2,000 romance novels.
The booksellers are fighting back the only way they know how — by protecting their livelihoods and questioning the motives of these corporate vultures. Good on them. If the tech giants want our cultural heritage, they can bloody well pay for it properly and maybe — just maybe — hire some actual human beings to do the reading instead of building glorified plagiarism machines.