Amazon Accused Of Scanning And Destroying Rare Books To Build AI Training Data

Amazon has been accused of scanning and destroying rare books as part of efforts to assemble datasets for training artificial intelligence models, according to an investigation by 404 Media.

The probe began when a bookseller grew suspicious of a large anonymous order placed on Biblio, an online marketplace that permits buyers to remain anonymous. To follow the shipment, the seller concealed an Apple AirTag inside one volume and tracked the package to a major Amazon facility in Nevada.

Workers at the site reportedly told the bookseller that employees in a unit called VGT3 remove book spines so pages can be scanned quickly, a process that destroys the original physical copies. An employee quoted by 404 Media wrote on an internal forum,

"I work at VGT3 here in Vegas, and all we do is scan books. Some are assigned to cut books, and others go to receive where they get books and scan the barcodes. We didn't have rates, but now we do, but it's not stressful."

The VGT3 unit is said to use a logo that depicts a T-Rex holding a book — an image 404 Media noted appears almost comically on-the-nose given the reported activity.

Amazon provided a brief statement to 404 Media saying: "Amazon purchases books through commercial channels to help develop and improve the products and services our customers use." The company did not directly address reports that workers strip bindings and dispose of scanned copies.

Context: Anthropic Lawsuit And Legal Ruling

The spotlight on this practice widened amid litigation involving AI developer Anthropic, whose filings revealed it used industrial equipment to cut and scan millions of books, then dispose of the originals. In that case a judge found that converting physical books into digital form and discarding the originals was "transformative," concluding it did not constitute copyright infringement under the court's reading — a ruling that alarmed authors and booksellers.

After those disclosures, booksellers said they increasingly suspected anonymous bulk orders were being funneled to AI firms. Sellers pointed to giveaway signs such as unusually large order sizes, seemingly random assortments of titles, and the presence of ISBNs on shipped items.

404 Media also reported that workers at the Nevada site appear to scan the ISBNs of every book they process. That practice lends weight to a worrying theory among sellers: that some firms may be working through lists of ISBNs to systematically digitize printed titles worldwide.

Why Sellers Are Alarmed

The books tracked in the investigation were described as rare — not always expensive first editions, but copies that are scarce in circulation. Sellers emphasized that books carry many kinds of value beyond monetary worth, including historical, intellectual, and sentimental significance.

"There are different types of value," one bookseller told 404 Media. "There's monetary value, obviously, but there are a lot of other types of value. There's historical value, intellectual value, sentimental value. All sorts of things, and all of those the AI companies don't care about. They just want the content as a bunch of words strung together."

Preservation advocates, authors, and sellers say the practice — if widespread — risks eroding cultural heritage and undermining trust between sellers and buyers. The episode has amplified calls for greater transparency from tech companies about their data-acquisition methods and for clearer legal standards around digitization and copyright.

Implications: The allegations raise legal, ethical, and cultural questions about how physical books are treated in the rush to build large AI models. Observers say the situation highlights the need for industry transparency, stronger protections for rare materials, and updated legal frameworks to address mass digitization.