Sep 2, 2026 · 41 min
Amazon’s book destruction exposes AI’s hidden supply chain
We Spoke to an Amazon Worker Destroying Books for AI
The episode connects the physical waste of AI training with a quieter threat: synthetic research entering scholarly databases and feeding future models.
- 1Amazon’s Las Vegas operation strips and scans books for AI training, then discards the originals beyond practical recovery.
- 2A tracked book reportedly reached a Mexican paper mill, suggesting destroyed AI-training material may re-enter circulation as paper.
- 3Fabricated authors and papers can move from open repositories into scholarly databases, creating feedback loops associated with model collapse.
Don't miss
A tracking device placed in a book reportedly followed its discarded remains to a Mexican paper mill.
The brief
At an Amazon warehouse in Las Vegas, workers reportedly remove book spines, scan pages for AI training data, and discard the physical originals.
The operation handles library books, foreign-language collections, and specialized texts, raising questions about what AI companies value—and what gets destroyed without scanning.
A bookseller hid a tracking device in a book and followed it to a Mexican paper mill, suggesting some discarded volumes may return as paper products.
The episode then turns to recurring fictional names and fabricated authors in academic material, showing how unverified repositories can push synthetic records into scholarly databases.
Together, the stories describe two feedback loops: physical books become disposable training inputs, while synthetic scholarship can become training data for future models.
Becky FerreiraMentioned
Amazon.com, Inc.
Las Vegas
404 MediaListen to the full episode and explore every guest, topic, and moment on PodLume.