Investigative Report Finds Amazon Buying and Destroying Rare Physical Books for AI Training

An investigative report published today by Ars Technica and 404 Media reveals that Amazon and affiliated data vendors have been systematically purchasing rare, out-of-print, and specialized physical books from independent sellers to digitize them for foundation model training.

According to internal sources and tracking logs, facility workers use high-speed automated destructive scanners—slicing off book spines to feed individual pages into optical scanning feeds—after which the physical volumes are shredded and discarded.

The revelation has sparked intense outrage among librarians, literary historians, and copyright attorneys, who condemn the physical destruction of cultural artifacts and question the legality of unauthorized scanning for commercial AI model building.