AI Book Shredding

Charlie Becker’s family bookstore in Houston has been connecting people with literature for more than 30 years.

When he was 12 years old, Becker remembers his dad making one of the store’s first larger purchases. Another local business planned to get rid of its collection of books, but first called Becker’s father.

“It was in their [company’s] last days, and they said my dad had to pick up the books. I went with him to the warehouse,” he said.

Becker grew up with that story of rescuing books bound for a landfill. Over the years, he has watched the entire industry of acquiring and selling books change with the rise of the internet, digital cataloging, and the emergence of major sellers such as Amazon.

For generations, booksellers have worked to preserve humanity’s writing. However, the recent revelation that millions of print books are being scanned to train artificial intelligence models, then destroyed presents an unprecedented challenge.

To make matters worse, book collectors and sellers say there’s no easy way to make it stop.

The practice of what has been dubbed AI “book shredding” burst onto the scene after a 2025 court document revealed that AI tech giant Anthropic purchased millions of printed books, removed their bindings, then scanned each page into digital datasets. Afterward, Anthropic shredded and discarded the originals.

It was revealed that the project was part of an ongoing expansion of Anthropic’s central library, which has an aim to collect “all the books in the world” and retain them “forever.” The undertaking was called Project Panama.

The same document noted that Tom Turvey, former head of partnerships for Google’s own book-scanning project, was hired to acquire material for the project. Turvey’s team emailed “major book distributors and retailers about bulk purchasing their print copies for Anthropic’s ‘research library.’”

Court filings from the Bartz v. Anthropic lawsuit, unsealed in January, named retailers such as Better World Books and World of Books as vendors from which Anthropic acquired thousands of books.

Anthropic did not respond to a request for comment.

“Sourcing books is a widely used approach for training large language models across the AI industry,” an Anthropic spokesman told technology website Tom’s Guide. “None of our data acquisition programs buy and destroy rare or antiquarian books.”

While a federal judge ruled the destructive scanning of legally purchased books qualified as transformative fair use—alteration of an original work for a new purpose—the practice has sparked a growing wave of public outrage.

Concern is also growing among book dealers, many of whom say systematically shredding books means losing not just words, but also cultural artifacts.

Suspicious Orders

One of the greatest challenges to stopping this practice is a lack of transparency around who is buying the large volumes of books.

In July, a 404 Media report flagged the book database ISBNdb for promoting print book acquisition services that would keep buyer information confidential.

ISBNdb has since changed the landing page on its website titled “Printed Books Sourcing for Your AI LLMs Dataset Needs.” The website now states that the company was “exploring demand” and has “chosen to pivot away from that direction.”

When questioned about this practice, a representative from ISBNdb reiterated the statement on the company’s website.

ISBNdb has never “purchased, scanned, or destroyed a book for AI training or anything else,” the representative told The Epoch Times. “We have never bought or sold printed books for AI training: no orders, no purchases, no books.”

But even if they had, that’s just one aspect of the issue. Nondisclosure agreements can be involved when sellers are working with big buyers, Susan Benne, executive director of the Antiquarian Booksellers’ Association of America, told The Epoch Times.

“It’s not currently common practice to ask a seller, ‘What are you going to use this book for?’”

The lack of available information is why Becker thinks it’s important to know what books are being scanned for AI use and then destroyed.

“Literally no one knows; that’s part of the problem,“ he said. ”Somebody who cares about our literary heritage should be in that pipeline somewhere, but that’s not what’s happening.”

—Autumn Spredemann; Stacy Robinson

BOOKMARKS

The U.S. Supreme Court ruled 5-4 on Monday that construction of President Donald Trump’s White House ballroom project can continue while the administration contests a lower court order that would block much of the development. The dissenting justices said Congress “explicitly prohibited” such construction without its approval.

Supreme Court Justice Samuel Alito recently disclosed that he made more than $30,000 in book sales and traveled to several countries with expenses paid by multiple organizations in 2025. The new filing was released after Alito said he would stay on the court and referred to justices’ time on the bench as “life tenure.”

Italy said Monday that it would suspend its open-border arrangements with Spain for another 15 days, and Spain responded by doing the same for Italy. The tit-for-tat comes amid tension  after thousands of young men crossed from Morocco into the tiny Spanish territory of Ceuta on July 30 and July 31.

NATO military activity in the Arctic poses a direct threat to Russia’s security and could lead to disaster, Russian Foreign Minister Sergei Lavrov wrote in an article published Monday. He said that such activity increases the “risk of incidents that could trigger an armed confrontation with potentially disastrous consequences.”

Sweden will help Finland with aerial surveillance until at least the end of the year, following accidental Ukrainian drone incursions into its territory. Both countries recently joined NATO after the Ukraine-Russia war began.

—Stacy Robinson