Amazon has been purchasing large quantities of rare books, removing their spines, and scanning their contents to use as training data for AI models, according to a report from 404 Media, which tracked a shipment using a hidden tracking device that led to an Amazon facility in Las Vegas. The facility, internally identified as VGT3, is marked with a symbol depicting a dinosaur holding a book. In response to the report, Amazon said it purchases books through standard commercial channels to help improve the products and services it offers customers.
The practice reflects a broader industry push to find fresh, high-quality text sources as AI companies exhaust readily available material from the internet, having already trained extensively on digitized web content and, in some cases, unauthorized book collections, as seen in a related copyright settlement involving Anthropic. Rare and out-of-print books are particularly valuable to AI developers because they predate the rise of AI-generated writing, reducing the risk of “model collapse,” a phenomenon in which AI systems trained on too much AI-generated content gradually decline in output quality.
