Inside Amazon's VGT3 Warehouse: Destructive Book Scanning for AI Training

Amazon operates a facility in Las Vegas, Nevada, known as VGT3, where books are destructively scanned for AI training data. [1] V VGT3 is housed within the same complex as LAS8, which handles Amazon's print-on-demand business. [1]

At the VGT3 warehouse, employees receive shipments of books and cut their spines off. [1]

After severing the spines, workers scan the loose pages and throw the remaining paper into a large shuttle. [1]

The existence of this destructive scanning process was uncovered when journalists from 404 Media placed a tracking device inside a shipment of rare books that a bookseller suspected was being acquired by an AI company. [1]

The bookseller suspected an anonymous AI company was acquiring the texts. [1]

The tracked shipment traveled across the country before arriving at the VGT3 facility. [1]

Online, Amazon employees who worked at the location have also described this process of receiving, cutting, and scanning the books. [1]

The physical destruction of rare books highlights the aggressive measures tech companies will deploy to acquire fresh, high-quality text data for model training. While scraping digital libraries has drawn copyright scrutiny across the industry, physical book scanning moves the ingestion process offline, introducing a labor-intensive hardware pipeline to digitize previously un-scraped works. What the sources don't address: Which specific Amazon AI models are being trained on this scanned data, and the total volume of books that have been destroyed through this pipeline.