Shipment of rare books tracked to an Amazon AI training facility

Summary

Simon Willison reports on a study by 404 Media; this research tracked the shipment of roughly 1,000 rare books to Amazon's VGT3 AI training facility in Las Vegas using an AirTag.

Why it matters

The study provides concrete evidence of how large-scale book scanning for AI training is carried out, highlighting the opaque nature of data collection by major tech firms. It raises concerns about rare book markets, intellectual property rights, and the environmental costs of destructive scanning processes.

Highlights

• A bookseller placed a massive order for 1,000 rare books on the Biblio marketplace without regard to price.
• An AirTag placed in one of the books tracked its delivery to Amazon's LAS8 VGT3 facility, confirming a direct link between rare book sales and AI data pipelines.
• Amazon employees on online forums confirmed that VGT3 scans large volumes of books in a destructive manner.

Critical take

While the AirTag tracking verifies a physical destination, the article lacks detail on how Amazon selects which books to scan and whether any protection or compensation measures are provided for rare book sellers.