Journalists Track Rare Book Shipment to Amazon AI Training Facility
Investigative journalists from 404 Media tracked a bulk order of approximately 1,000 rare books using an Apple AirTag hidden inside one of the volumes. The shipment was traced directly to an Amazon facility in Las Vegas, where online forum discussions by workers confirmed that books are destructively scanned. This investigation provides concrete physical evidence of how tech giants acquire copyrighted and rare physical books to build training datasets for AI models. It highlights the controversial and aggressive sourcing methods used by AI companies, bypassing traditional licensing agreements. The AirTag tracked the shipment to the VGT3 section of Amazon's LAS8 facility in Las Vegas, which features a logo of a dinosaur tearing into a book. Online discussions among Amazon employees confirmed that this specific location is used for the destructive scanning of large volumes of books.
## BACKGROUND
AI developers require massive amounts of text data, including books, to train large language models. While digital text is often scraped from the web, physical books represent a valuable source of high-quality, copyrighted material that is not easily accessible online. Destructive scanning refers to the process of cutting the spines off books to feed pages quickly into high-speed scanners.