Amazon Is Buying and Scanning Rare Books for AI Training, Report Says

Amazon is purchasing rare and out-of-print books and scanning them for artificial intelligence training, according to a 404 Media investigation that tracked one shipment to an Amazon facility in Las Vegas.

The report said Amazon removes the spines from some books before scanning their pages, leaving the physical copies unusable. 404 Media placed a tracking device inside a rare book and followed it to a facility identified as VGT3.

"Amazon purchases books through commercial channels to improve the products and services customers use," the company told 404 Media.

Rare books offer hard-to-find training data

Rare and out-of-print books can contain text unavailable online, making them potentially valuable to companies building large language model datasets. Many older books also predate the widespread use of generative AI, giving developers access to human-written material that another model did not produce.

Researchers have warned that repeatedly training AI systems on generated content can contribute to model collapse, where output quality and diversity deteriorate over time.

The report does not establish how many books Amazon has purchased or destroyed, or which AI products have used the resulting text.