Amazon's Secret Operation: Rare Books Destroyed to Train AI
My analysis of recent events has revealed a troubling trend: at one Amazon facility in Las Vegas, a veritable "data factory" is unfolding, where valuable printed publications are systematically stripped of their bindings, scanned, and sent to waste. This is not just a logistical process—it is an industrial sacrifice of book culture for the sake of artificial intelligence progress.
How the scheme was uncovered
The starting point was anomalous orders from used-book dealers. Sellers recorded wholesale purchases of hundreds of diverse books, including rare copies, which is atypical for private collectors. In July, one store received an order for approximately 1,000 books through the Biblio marketplace from an anonymous buyer. An Apple AirTag tracker was embedded in one of the publications, allowing the shipment's path to be traced.
The signal led to the VGT3 unit in North Las Vegas, which specializes in processing printed materials. After reviewing employee posts and conducting interviews with former workers, I confirmed that the facility regularly receives large batches of publications.
The destruction technology
The process looks like this: first, copies are registered and barcodes or ISBNs are scanned, then bindings are cut off to separate them into individual sheets. After that, the pages are loaded into high-speed scanners. This approach radically speeds up digitization but renders the book physically unfit for restoration.
Employees describe this as the unit's primary task: receiving batches, inventorying, cutting, and scanning. In early 2026, there were supply disruptions, but the flow has now been restored. Amazon has not publicized this activity, but in response to inquiries, it confirmed that it purchases printed publications for "developing and improving products and services," without disclosing the scale and target areas.
My comment: This practice is a glaring symptom of the race for high-quality data. While some companies rely on open datasets, Amazon invests in physical infrastructure, destroying cultural heritage to train models. The question is not about efficiency but about ethics: how many rare books are we willing to sacrifice so that AI becomes "smarter"? The market for such data is clearly undervalued, and this sets a dangerous precedent for the entire industry.