CONSUMING MASS QUANTITIES... OF BOOKS

PREV ARTICLE NEXT ARTICLE FULL ISSUE PREV FULL ISSUE

V29 2026 INDEX E-SYLUM ARCHIVE

The E-Sylum: Volume 29, Number 34, 2026, Article 22

Last week we discussed the mysterious bulk orders used and rare bookdealers have been receiving, suspecting that AI labs are scanning them for training data to improve their models. Well, a curious journalist put a tracking device in a rare book, and this is what they learned. -Editor

Amazon is buying massive quantities of books, scanning them for AI training data, and destroying them in the process.

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.

That final destination was an Amazon warehouse in Las Vegas, Nevada. 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. 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.

"Amazon purchases books through commercial channels to help develop and improve the products and services our customers use," an Amazon spokesperson told me in a statement.

The world's AI companies are constantly looking for, and spending extreme resources to locate, more material to train their AI models. With books, that sometimes means destroying them in the process, something that large parts of the public have spoken up against, and which we can now confirm Amazon is doing.

In July, I published a story about booksellers who reported a historical spike in sales starting in the past year. They suspected this spike in sales was due to AI companies acquiring any books they can in search of new training data. Printed books are valuable as training data because a lot of the text they contain is not readily available on the internet, which AI companies have already scraped.

"I work at VGT3 here in Vegas, and all we do is scan books," one Amazon employee wrote on a forum for Amazon workers. "Some are assigned to cut books, and others go to receive where they get books and scan the bar codes. We didn't have rates, but now we do, but it's not stressful."

"Working at VGT3 is nice all we do is scan books," another Amazon employee wrote. "It's so cool."

The scanning crews record the ISBN of each book scanned. That makes me think that older books from before ISBNs were used are immune, at least for now. But is that book-munching dinosaur coming for our numismatic literature, too?

An E-Sylum reader passed along a very detailed Pulitzer-worthy article about this. Here's an excerpt - see the complete article online. This is industrial digitization at a boggling scale. -Editor

Millions of books are being destroyed to train AI? Nope, it's tens of millions.

News reports have centered on Anthropic's book scanning because details of its program, called Project Panama, came to light in court filings. The company's goal was to "destructively scan all the books in the world." However, it is far from the only company involved. The book scanning ecosystems includes AI developers, book acquisition firms, and scanning companies.

In a statement to The Guardian, Anthropic implied that destructive scanning was standard operating procedure, saying, "sourcing books is a widely used approach for training large language models across the AI industry."

In 2024 and 2025, the AI scanning projects mostly happened under the radar. Everyone involved signed NDAs and the companies primarily bought from mega-sellers, like Better World Books, World of Books, and Thrift Books. Individual bookstores, even Powell's and The Strand, were considered too small.

Eventually, the need for books, particularly specialized titles, outgrew the mega-sellers, and orders began to flow last December through the used book marketplaces, beginning with Alibris before spreading to Biblio.com, Abebooks, and Amazon's Marketplace.

As far as I can tell, no news outlet has reported on ARC Document Solutions, which is scanning as many books every month as Anthropic's Project Panama envisioned scanning in its entirety.

ARC's current rate of scanning of 2 million books per month is up from the 1.5 million books claimed on a trade show banner six weeks ago, at the beginning of July.

In a promotional video, an ARC executive says that many of the company's scanning locations "run twenty-four hours a day, three shifts, including weekends," adding, "It's very exciting for us."

A sign above a monitor at one of their scanning facilities appears to set a production goal of 200 books per day. For a roomful of scanners, that would be about 3,000 books per shift. In a year, one group of scanners working seven days a week can process more than two million books.

Suddenly, ARC's two-million-books-per-month claim doesn't seem so far fetched. Do the math. That's 20+ million books per year, and we are in the second or third year of AI book scanning. ARC is just one company. The total number of books destructively scanned to feed AI could easily be more than 50 million.

Just two weeks ago, a story on book scanning was headlined, "Someone Is Mysteriously Snapping Up Used Books." That already seems quaint. It's more like "Every Serious AI Company Is Buying All the Books It Can Find."

To read the complete article, see:
All the Books in the World (https://downtownbrown.substack.com/p/five-fallacies-ai-and-destructive-scanning-of-books)

In the Soylent-Green-is-People department, some of those destructively scanned books may be turned into toilet paper in Mexico. -Editor

A bookseller I know received an order for a book to be sent to a Trifinity warehouse in Kenosha, Wisconsin, a common destination for AI company orders. My friend put a tracking device in the book and watched as it went to Kenosha, then Las Vegas, and finally to a warehouse in Mexicali, Mexico. The warehouse is part of a large Papel San Francisco paper plant.

The plant is a leader in manufacturing post-consumer recycled products, including toilet paper.

If there is any remaining doubt that AI is destructively scanning books, I hope that this ends it. There is no plausible reason for a company to pay for a book and then send it to a paper recycling plant unless it has been cut up and turned into loose paper during a scanning process.

Books are, in the words of AI company engineers, perhaps the most valuable data for training their large language models.

And now we learn that at least one AI companies is trucking their destroyed books thousands of miles and disposing them in a foreign country. The brave bookseller who put the tracking device in the book doesn't want to be named because the AI companies have bottomless resources to pursue their critics.