Amazon Employees Are Literally Cutting the Bindings off Rare Books All Day Long to Train Language Models

Amazon employees at a Las Vegas warehouse are allegedly cutting the bindings off rare books to scan their pages for training AI language models, according to an investigation by an independent technology outlet that tracked a book with a hidden device to the facility. The process destroys the physical copies, and Amazon has not confirmed the books are used for AI training, though it said it 'purchases books through commercial channels to improve the products and services that customers use daily.' The practice highlights the growing demand for training data and follows Anthropic's $1.5 billion settlement over the use of pirated books in training its Claude model.

Amazon Employees Are Literally Cutting the Bindings off Rare Books All Day Long to Train Language Models The relentless pursuit of artificial intelligence is silently sacrificing our irreplaceable printed heritage for pristine algorithmic data Online retail giant Amazon is allegedly destroying thousands of rare books at a Las Vegas warehouse to scan their pages and train artificial intelligence language models, according to a new investigation. The news came after an independent technology outlet placed a tracking device inside a rare book https://www.ibtimes.co.uk/amazon-book-scanning-ai-training-controversy-1814811 it suspected was being purchased for AI training purposes and followed its journey across the US to a Nevada facility. Once the shipment of obscure volumes arrived at the VGT3 warehouse, workers reportedly removed the bindings from the books so the loose pages could be processed through high-speed scanners. The physical copies are destroyed during the digitisation process, leaving behind discarded paper and digital files that investigators claim are used as training material for Amazon's AI systems. Employees at the Las Vegas location told investigators that their daily work involves receiving large commercial shipments of printed books and removing their spines to prepare them for scanning. The internal logo for the Amazon team operating inside the warehouse reportedly features a dinosaur holding a book, an unusual image given the alleged process taking place inside the facility. Amazon has not publicly confirmed that books processed at the warehouse are destroyed specifically for AI training. The Hunt for Human Data Technology companies investing heavily in artificial intelligence require enormous quantities of text to train large language models, with many firms facing challenges over the availability and legality of training data. Rare books that are out of print or unavailable through digital archives have become a potential source of text for companies developing AI systems. Older literature is often considered valuable training material because it predates the widespread use of generative AI tools, reducing concerns that models could be trained on synthetic content. AI developers have raised concerns about training models on excessive amounts of computer-generated material, a phenomenon sometimes referred to as 'model collapse.' Researchers have warned that repeatedly training AI systems on AI-generated content could reduce output quality over time, potentially leading to less accurate or less diverse results. A Controversial Industry Practice Amazon is not the only technology company to face scrutiny over how it obtains books and other written material for AI development. AI company Anthropic has also faced legal challenges https://www.ibtimes.co.uk/anthropic-global-watermarking-ai-content-compliance-1814453 over its use of copyrighted material in training its Claude language model. The company reached a settlement in a lawsuit brought by authors, agreeing to pay up to $1.5 billion £1.1 billion in damages over allegations involving the use of pirated books for AI training. Anthropic has denied wrongdoing and maintained that its use of training material was connected to developing AI systems. Despite growing legal disputes over copyright and artificial intelligence, companies have provided limited public detail about exactly how they source training materials. In a statement regarding its warehouse operations, Amazon said it 'purchases books through commercial channels to improve the products and services that customers use daily.' Booksellers and archivists are now questioning whether increased demand for rare and out-of-print books could be linked to AI companies seeking new sources of training data. The possibility that valuable physical books are being acquired primarily for digitisation has raised concerns among preservationists and collectors who fear rare works could be lost in the process. The demand for high-quality written data is expected to continue growing as technology companies compete to develop increasingly advanced AI models. As companies search for new sources of human-created text, the debate over copyright, preservation and the future of physical books is likely to intensify. IBTimes UK cannot independently verify claims regarding Amazon's warehouse operations or the alleged destruction of books for AI training. The allegations are based on an external investigation and should be treated accordingly. © Copyright IBTimes 2025. All rights reserved.