Silicon Valley tech firms are purchasing thousands of second-hand books from British booksellers to feed AI chatbots before destroying the physical volumes.
The practice, known as destructive scanning, involves slicing the pages from book bindings, scanning them into industrial machines to train AI models such as Anthropic's Claude and OpenAI's ChatGPT, and discarding the original paper.
Lucy Hingston, 49, owner of Compass Books in Dartmouth, Devon, grew suspicious earlier this summer after receiving a bulk order for books covering niche topics including Western Peru geology, Cornish pioneers in South Africa, the defunct Eastern Counties Railway, cheeses of the world, and Hong Kong jewellery.
The buyer listed a shipping address based at Heathrow Airport. After contacting other independent booksellers across the UK, Hingston discovered that her shop was not the only one receiving such orders.
"It's a moral dilemma," Hingston said. "They're my books, my babies. I feel like we've been eaten by the AI monster. It's difficult. I want the sale, but not to somebody who will b****r up everybody's books."
Hingston noted that some of the titles purchased from her store were collectable and valuable, raising concerns that rare or out-of-print books could be destroyed.
Dan Conway, chief executive officer of the Publishers Association, condemned the practice as part of a broader threat to creative industries.
"Bulk-buying books just to rip them apart to scan is shocking, but it's just the tip of the iceberg when it comes to the brazen theft of copyright materials by the tech giants," Conway said.
"They're not just destroying books," Conway added. "Through their shameless actions, including the illegal use of content from books and journals without payment or permission, they are harming our world-leading creative industries and the livelihoods of authors, illustrators and journalists. This behaviour needs to stop."
Last year, the Daily Mail launched its Make It Fair campaign to stop tech companies from taking copyrighted material without compensation. The campaign has received backing from prominent figures including musician Sir Elton John, television producer Simon Cowell, and Universal Music chief executive Sir Lucian Grainge.
Secret Scanning Projects and Legal Loopholes
Court documents released earlier this year revealed that Anthropic, the AI company behind the Claude chatbot, initiated destructive scanning in early 2024 under a secret initiative named Project Panama.
"We don't want it to be known that we are working on this," an Anthropic employee declared in an internal memorandum included in the court filings.
The filings showed Anthropic spent millions of dollars to buy millions of print books, stripped their bindings, cut their pages to size, and scanned them into digital formats before discarding the paper originals.
Anthropic previously trained Claude on more than seven million digitised books obtained from illegal pirate websites, including five million titles taken from a single pirate repository.
Earlier this month, a federal judge in San Francisco approved a $1.5 billion (£1.1 billion) out-of-court settlement in a 2024 class-action lawsuit brought against Anthropic by three authors in California.
The lawsuit prompted Anthropic to switch to purchasing physical books, allowing the company to avoid copyright infringement claims under the American legal principle known as the first sale doctrine. Established by the US Supreme Court in 1908, the doctrine allows buyers of physical goods to dispose of or modify their property as they choose.
Purchasing existing physical books offered tech companies a cheaper alternative to paying writers to create high-quality training text. According to specialized scanning firm eRecordsUSA, cutting book spines allows for significantly faster processing and guarantees shadow-free scans.
Anthropic is not the only firm engaging in mass scanning. An investigation by news outlet 404Media revealed that Amazon uses a facility designated VGT3 outside Las Vegas to scan large volumes of second-hand books.
"I work at VGT3 here in Vegas, and all we do is scan books," an anonymous Amazon worker wrote on an online forum. "Some are assigned to cut books, and others go to receive where they get books and scan the bar codes."
Pre-2022 Books and Model Collapse
Tech firms specifically target second-hand booksellers with online catalogues, seeking out titles published before 2022. That year marked the widespread popularity of generative AI tools following the launch of OpenAI's ChatGPT.
Books published prior to 2022 are guaranteed to be free of AI-generated text. Feeding AI-generated content back into training systems causes a technical failure known as model collapse.
"It's the AI equivalent of a feedback loop gone wrong," explained Yarin Gal, associate professor of computer learning at Oxford University. "The more models feed on their own output, the further they drift from reality. Model collapse threatens to create an AI echo chamber."
Booksellers across Britain and abroad have reported suspicious bulk purchases. Finn Windle, 30, who works at The Minster Gate Bookshop in York, described an unusual batch order in May involving diverse non-fiction topics such as local history and geology.
"We had an unusual batch order in May, quite a lot of books, and none of them were related subjects," Windle said. "It was quite confusing. There was local history, geology, a real mix of things, academic books and books for the general reader. Generally, non-fiction."
The purchaser was an American entity named Zoom Books, which placed two separate orders. Similar orders from Zoom Books were reported by booksellers in Germany, New Zealand, and Australia.
"If we knew destructive scanning is what they were doing, we wouldn't have authorised the order," Windle added. "We're not pleased about it, it's awful. Some of these titles aren't even in print any more and there are no other copies available."
In Felixstowe, Suffolk, 64-year-old Helen Bott of Treasure Chest Books received initial orders for small batches in early May from Zoom Books, followed a week later by a massive order for 340 books.
Bott, whose shop specializes in transport and railway titles, noted that half the order consisted of railway books alongside topics such as Latin, school history, shipping companies, and Devon village history.
"We noticed these were all non-fiction," Bott said. "It felt like they were looking for unusual books: a history of a village in Devon, for example. You don't want to see books destroyed. But I'm conflicted about it. From a business point of view, it's good. But destroying books, that is the last thing we want to see."
Industry Divide and Corporate Responses
Meryl Halls, managing director of the Booksellers Association, described the bulk purchases as a concerning development.
"The potential scenario of tech companies bulk-ordering second-hand books to train their large language models is a concerning development," Halls said. "If true, it is another example of the lengths some AI companies may go to in order to access copyright-protected material without transparency or compensation."
Some booksellers viewed the trend differently. Patrick Kelly of Book Mongers in Brixton, south London, welcomed the potential business.
"We have been joking and hoping that that would happen to us!" Kelly said. "We could use the business. My suspicion is that you need a catalogue of your inventory, something we don't have. To be honest, I don't understand why the shops affected are complaining."
The practice has sparked wider debate, prompting Elon Musk to state on social media platform X last month that he instructed his AI team to preserve rare volumes and scan them by manually turning pages rather than cutting off bindings.
In response to inquiries, an Anthropic spokesperson stated: "Claude is trained on a mix of publicly available web data, commercially acquired datasets, and data we generate ourselves. Sourcing books is a widely used approach for training large language models across the AI industry. None of our data acquisition programs buy and destroy rare or antiquarian books."
Zoom Books denied involvement in destructive scanning. An Amazon spokesperson stated: "Amazon purchases books through commercial channels to help develop and improve the products and services our customers use."
Mass destruction of literature has historical precedents, including the 1933 book burnings in Nazi Germany across Third Reich university cities and the Renaissance Bonfire of the Vanities led by friar Girolamo Savonarola in Florence. Today, independent booksellers remain divided as tech companies seek pre-digital texts to expand their AI capabilities.
