Amazon Destroys Rare Books for AI Training Data

Amazon has been purchasing and destroying rare books to generate training data for its artificial intelligence models, according to an investigation that tracked a shipment of books to a company warehouse in Las Vegas. The practice, which involves cutting the spines off books to scan their pages, has sparked concerns about the loss of irreplaceable texts and the ethics of using physical books for AI development.

The investigation began when a bookseller, who wished to remain anonymous, received a bulk order for approximately 1,000 books on the online marketplace Biblio. Suspecting the buyer was an AI company, the seller agreed to place an Apple AirTag inside one of the volumes. The device tracked the shipment from a California airport to Milwaukee, then by truck to an Amazon warehouse complex in Las Vegas. The book ultimately arrived at a facility designated VGT3, which is part of a larger Amazon distribution center.

Amazon employees who work at the VGT3 site described their duties on internal forums. "I work at VGT3 here in Vegas, and all we do is scan books," one employee wrote. "Some are assigned to cut books, and others go to receive where they get books and scan the bar...