Amazon destroys rare books to train AI models, investigation finds

Amazon is purchasing bulk lots of rare books and systematically destroying them after scanning their pages to train its frontier AI models. The practice was uncovered through an investigation by 404 Media, which tracked the fate of rare volumes to a specialized facility in Nevada.

The discovery began when a bookseller planted an AirTag inside a rare book, which tracked the shipment directly to Amazon's VGT3 AI training facility in Las Vegas. Reports indicate that workers at the site are tasked with tearing books from their spines to facilitate high-speed scanning. In a detail that underscores the nature of the operation, workers at VGT3 reportedly use a logo featuring a T. rex devouring a book.

The race for unique data

This operation is part of a broader industry scramble for high-quality training data. As AI companies race to build "frontier models," they require massive amounts of unique, non-digitized text to improve model performance and diversity. Some firms appear to be targeting books with ISBN numbers in an apparent effort to methodically scan every printed book in existence. This strategy focuses on obscure works that may have low monetary value but possess high intellectual or historical significance.

Cultural heritage at risk

The practice has sparked severe ethical concerns regarding the permanent loss of physical cultural heritage. Critics argue that destroying the only surviving copies of specific editions to create proprietary AI training sets is a dystopian trade-off. The primary concern is that while the original physical artifacts are obliterated, the resulting knowledge is often locked behind corporate paywalls or subject to AI censorship, removing the information from the public commons entirely.

Amazon's response

In an official statement, Amazon defended its procurement process, noting that the company "purchases books through commercial channels to help develop and improve the products and services our customers use." The company did not specifically address the reports of the destructive scanning process or the disposal of the physical volumes after digitization.

What remains to be seen

As the investigation expands, the industry is watching to see if other AI giants are employing similar destructive tactics to secure a data advantage. The revelation has already drawn sharp criticism from high-profile figures; investor Michael Burry described the practice as "evil incarnate." Whether this will lead to new regulations regarding the preservation of physical archives in the age of generative AI remains an open question.