AI News

Amazon rare books cut up for AI training

#Amazon#Training data#Copyright

How 404 Media traced Amazon rare books to Las Vegas

The story begins with one rare book and a tracking device. Reporters at 404 Media published what they found on August 17, 2026.

A tracker hidden in a rare book led 404 Media to VGT3

404 Media placed a tracking device inside a rare book it suspected an AI company would acquire, and followed it to its final stop. The outlet says this revealed Amazon's book-buying operation, which had not been reported before.

The top of the 404 Media article “We Tracked a Shipment of Rare Books. It Ended at an Amazon AI Training Facility,” published August 17, 2026
404 Media's August 17, 2026 report traced where the rare books went.404 Media — We Tracked a Shipment of Rare Books. It Ended at an Amazon AI Training Facility
"A 404 Media investigation was able to reveal Amazon’s book buying operation, which hasn’t been previously reported, by placing a tracking device in a rare book we suspected would be acquired by an AI company for training data, and following it around the country to its final destination."— from 404 Media's investigation
View official source →

VGT3, the Amazon team at the Las Vegas warehouse

The tracker ended up at an Amazon warehouse in Las Vegas, Nevada, and the Amazon team that works there is called VGT3. Its logo shows a dinosaur with bared teeth and a book in its hands.

"That final destination was an Amazon warehouse in Las Vegas, Nevada." (the destination)／"The logo of the Amazon team that works at this warehouse, called VGT3, is a dinosaur, brandishing its teeth and with a book in its hands." (the VGT3 team)— from 404 Media's investigation
View official source →

Las Vegas warehouse staff cut bindings off for book scanning

Everything in this section comes from the employees 404 Media spoke to, not from Amazon. They say the job at the Las Vegas site is receiving large shipments of printed books and cutting the bindings off, and the printed book does not survive. The reason they give is speed, since a book without its binding goes through a scanner faster.

"Amazon employees who work at this location say all they do is receive massive shipments of printed books which they then cut the bindings off in order to scan the books more quickly. The printed book is destroyed in the process."— from 404 Media's investigation
View official source →

What model collapse research says about AI training data

In May 2023, Ilia Shumailov and colleagues reported that training on content generated by other models causes defects that cannot be undone. They named the effect model collapse. The rare parts of the original data are the first to vanish, and the paper shows it can happen in large language models (LLMs, the kind of AI behind chatbots).

The authors also draw a conclusion about data. As more model output mixes into what is crawled from the internet, they write, data on real people interacting with these systems becomes more valuable.

The top of the arXiv page for “The Curse of Recursion: Training on Generated Data Makes Models Forget,” showing the authors and the abstract
The abstract page of the paper that reported model collapse.arXiv — The Curse of Recursion: Training on Generated Data Makes Models Forget
We find that use of model-generated content in training causes irreversible defects in the resulting models, where tails of the original content distribution disappear. We refer to this effect as Model Collapse and show that it can occur in Variational Autoencoders, Gaussian Mixture Models and LLMs. ／ Indeed, the value of data collected about genuine human interactions with systems will be increasingly valuable in the presence of content generated by LLMs in data crawled from the Internet. — from the paper's abstract
View official source →

Amazon rare books and the rising value of human-made data

One tracked rare book ended up with VGT3, Amazon's team at a Las Vegas warehouse, where employees say bindings are cut off for scanning and the printed book is destroyed. Read alongside the model collapse paper, the trail suggests why such books could matter to AI developers: the more model output there is online, the more human-made data is worth. The free portion of 404 Media's report does not say whether that is Amazon's reason.

FAQ

How did the Amazon rare books story come to light?
404 Media put a tracking device in a rare book it expected an AI company to buy, then followed the shipment across the country. It ended at an Amazon warehouse in Las Vegas. 404 Media says Amazon's book-buying operation had not been reported before.
What do Amazon employees say happens to the books?
Employees told 404 Media that the work is receiving printed books and cutting the bindings off so the books scan faster. The printed book is destroyed in the process. That is the employees' description, not a statement from Amazon.
What is model collapse?
It is what happens when models keep training on content that other models generated: the resulting models develop defects that cannot be undone. The researchers who named it found that the rare parts of the original data disappear first, and that data on real people using these systems grows more valuable as model output spreads online.