A tracking device concealed inside a book order has revealed the fate of rare and used volumes purchased for artificial intelligence training: shipment to a Las Vegas facility where their bindings are stripped, their pages scanned, and the physical copies discarded.
How the Discovery Was Made
The revelation came after a tracker was planted inside a book order, allowing investigators to follow the shipment to its final destination. Rather than landing on a collector's shelf or a library reading room, the volumes arrived at an industrial facility in Las Vegas linked to Amazon's data operations.
At the site, books are systematically disassembled. Bindings are removed so that individual pages can be fed through scanners, converting the printed text into digital data. Once the scanning is complete, the physical copies are destroyed rather than preserved or resold.
A tracking device turned a routine book order into a window on how the printed word is being fed into the machine.
Books as Fuel for AI
The practice underscores the growing appetite of technology firms for high-quality text to train large language models. Books, with their edited prose and structured information, are prized sources of training material, and companies have been racing to acquire content in a variety of formats.
Scanning physical volumes offers a route to text that may not exist in easily accessible digital form, including rare and out-of-print titles. But the destruction of the originals has raised concerns among readers, collectors, and preservationists who worry that irreplaceable copies are being lost in the process.
The episode adds to a broader debate over how AI developers source their training data and what happens to the materials once they have served their purpose. Key tensions include:
- The value of rare and used books as cultural artifacts versus raw data
- The permanence of destroying physical copies after digitization
- Questions of transparency around how AI companies acquire content
As demand for training data continues to climb, the fate of these Las Vegas-bound books offers a stark illustration of the trade-offs being made in the rush to build ever more capable AI systems.