Secondhand booksellers across the UK and Ireland are reporting an unusual wave of bulk purchases from unidentified buyers, fueling speculation that artificial intelligence companies are quietly stockpiling printed books to use as training material.
Bookshops contacted by the Guardian said they had fielded orders from buyers in the US, Canada, continental Europe and the UK. Similar patterns have also been observed internationally, with dealers in the US, Australia and Europe describing comparable transactions.
Stuart Manley, co-owner of Barter Books in Alnwick, Northumberland, said orders that began arriving three months ago stood out because they lacked the thematic coherence of typical purchases. “Normally a book order is on a theme but this is all over the place,” he said. One recent order included an Estonian translation of John le Carré’s The Mission Song, a specific imprint of Anne Brontë’s Agnes Grey, and the October 1983 edition of Warship magazine. Manley said he had sold hundreds of random titles to three buyers for about £4,000.
Jim Shaughnessy of MW Books in Claregalway, Ireland, said large, varied orders had been arriving since May. “They are sporadic, presumably machine-driven and have little in terms of thematic currency,” he said, listing examples from books on 18th-century African agricultural implements to biographies of 1950s racing drivers. David Gower-Spence of BookLovers of Bath said he received a rush of orders between May and July for about 200 books from buyers also linked to other dealers. One buyer, Canada-based Zoom Books, which describes itself as “North America’s leading book recycling company,” has been contacted for comment.
An anonymous UK bookseller said random orders from different buyers sometimes arrived at the same address. A postcode used by several buyers points to a group of freight warehouses near Heathrow Airport, according to the seller, who has received orders for about 6,000 books since January worth thousands of pounds. The buyers were paying top prices and not seeking discounts, despite bulk requests. “It’s disrupting the secondhand bookselling business because this isn’t typical for the trade,” the seller said.
The spate of purchases has prompted speculation that AI firms intend to scan the books before pulping them. Manley said: “My theory is that it’s masses and masses of AI money being squandered.” Earlier this year, The Washington Post revealed that Anthropic, the developer of the Claude chatbot, had spent tens of millions of dollars buying books, slicing off their spines for scanning, and sending them for recycling. Anthropic said that sourcing books is a widely used approach for training large language models and that none of its data-acquisition programs buy and destroy rare or antiquarian books.
Last month, tech site 404 Media reported that a books database had suggested pre-2022 books were ideal AI training material because their text was not polluted by potential chatbot-generated content. The database, ISBNdb, said the webpage was a “test of market interest” and had been taken down. Booksellers’ online forums show mixed reactions, with some lamenting that volumes may be stripped and pulped for AI training and others noting that some requested titles, such as Gulliver’s Travels, are freely available in the public domain. Orders are typically placed through marketplaces such as Biblio and AbeBooks, both of which were contacted for comment.