Rare Books Shipment Tracked to Amazon AI Facility Where Books Are Allegedly Scanned and Destroyed

An investigation by 404 Media tracked a shipment of hard-to-find books to an Amazon facility in Las Vegas, where workers allegedly cut bindings and scanned pages, destroying the books. Amazon stated it purchases books through commercial channels to develop products and services, but the practice has raised concerns among booksellers and librarians about the use of physical books for AI training data.

Rare Books Shipment Tracked to Amazon AI Facility Where Books Are Allegedly Scanned and Destroyed An investigation reportedly traced a shipment of hard-to-find books to an Amazon facility in Las Vegas Amazon is facing fresh scrutiny over the use of physical books for artificial intelligence training after an investigation reportedly tracked a shipment of hard-to-find titles to a company facility in Las Vegas, where workers allegedly cut the bindings from books before scanning their pages and eventually destroying them. The investigation, reported by 404 Media, involved placing a tracking device inside a book from a large order and following its journey across the United States. The shipment ultimately reached an Amazon facility associated with an operation known as VGT3. The discovery has raised concerns among booksellers, librarians and readers about what happens to physical books once they are acquired as potential sources of AI training data https://www.ibtimes.co.uk/ai-companies-demand-printed-books-copyright-concerns-1810831 . Amazon Facility Reportedly Cuts Books Apart According to workers cited in the investigation, the Las Vegas operation receives large shipments of printed books. Employees reportedly remove the bindings, allowing the individual pages to be processed through scanners more quickly. The process means the physical books are effectively destroyed after their contents have been digitised. The facility's VGT3 operation reportedly uses a logo showing a dinosaur holding a book, an unusual detail that has attracted attention online following the investigation. Amazon https://www.ibtimes.co.uk/amazon-wishlist-privacy-concerns-1781706 has not publicly said that the books are being destroyed specifically to train a particular AI model. In a statement to 404 Media, the company said it 'purchases books through commercial channels to help develop and improve products and services our customers use.' That wording has nevertheless done little to quiet questions about why large quantities of physical books are being purchased and processed in this way. Why Are AI Companies Looking for Old Books? The growing interest in physical books is linked to a major problem facing AI developers https://www.ibtimes.co.uk/meta-ai-model-hacks-systems-security-test-1812825 : finding large quantities of high-quality text that models have not already consumed. Older and out-of-print books can be particularly attractive because they contain material that may not be readily available online. They also pre-date the explosion of generative AI, meaning they are unlikely to contain text produced by other AI systems. That matters because researchers have warned about the possibility of 'model collapse', a situation in which repeatedly training AI models on synthetic, AI-generated material can gradually reduce the quality and diversity of their output. The practice is not limited to Amazon. Investigations and court records have previously revealed that AI companies have acquired physical books in bulk and used destructive scanning techniques to digitise them. Anthropic, for example, has acknowledged purchasing books commercially, while court documents have detailed its use of physical books in its training-data efforts. Are Rare Books Actually Being Destroyed? There is an important distinction between genuinely valuable antiquarian books and books that are simply difficult to find. Some reports have described the books involved as 'rare,' but this does not necessarily mean priceless first editions or unique historical manuscripts are being fed into scanners. A fact-check published by Yahoo News noted that Anthropic has distinguished between genuinely rare or valuable books and harder-to-find professional, reference and academic titles. The company has also said that its programmes do not purchase and destroy rare or antiquarian books. Nevertheless, booksellers have raised concerns about the broader practice of destroying physical copies that may have historical or cultural value. The Guardian reported that booksellers in Australia had become increasingly concerned that old titles were entering an AI supply chain, particularly as demand for physical material from AI companies has increased. The Controversy Over AI and Human Knowledge The situation highlights a growing tension between technological development and the preservation of physical culture. For AI companies, scanning books can provide vast quantities of text that can potentially improve future systems. For collectors, historians and booksellers, however, a physical book can represent far more than the information contained in its pages. Once a scarce book has been cut apart and pulped, the original object cannot simply be recovered. © Copyright IBTimes 2025. All rights reserved.