AirTag Tracks Bulk Book Shipment to Amazon AI Scanning Facility in Las Vegas

An Apple AirTag hidden inside a bulk shipment of rare books tracked the physical location of a secret data collection operation to an Amazon facility in Nevada.

The hidden tracker traveled through commercial delivery channels until arriving at Amazon’s VGT3 facility in North Las Vegas. Tech news outlet 404 Media conducted the tracking operation after placing the device inside an order placed by buyers making large-volume book purchases.

The final stop was Amazon’s VGT3 team. Employees told the outlet all they do is receive massive shipments of printed books, then cut the bindings off so the pages feed a scanner more quickly. The book is destroyed in the process. The team’s extremely fitting logo is a dinosaur clutching a book in its claws.

Industrial paper cutters slice off the bound spines of physical books upon arrival at the facility. Removing the bindings allows high-speed sheet feeders to process thousands of loose pages per hour before the paper material is discarded.

Booksellers across the country reported an influx of buyers purchasing full inventories of physical volumes over the past year. These commercial buyers purchased titles indiscriminately across genres without requesting condition reports or rare-book provenance.

Physical texts printed prior to 2022 provide human-authored text free from synthetic outputs produced by current large language models. AI developers seek pre-2022 print material to avoid model collapse, a degradation in performance that occurs when artificial intelligence systems train on machine-generated data.

Amazon said it “purchases books through commercial channels to help develop and improve the products and services our customers use.”

Destructive scanning operations are also utilized elsewhere in the technology sector. AI developer Anthropic previously ran an internal program designated Project Panama, which digitized millions of physical books through destructive scanning to compile training datasets.

Federal courts continue to review copyright disputes concerning the digitization of physical books for machine learning models. A federal judge ruled that scanning legally acquired print books for model training constitutes fair use, while separate copyright infringement lawsuits involving digitized print catalogs remain pending.