Today, Amazon is one of the world's biggest tech companies, with vast cloud computing and AI operations as well as its mammoth global online retail platforms. But its brand still has a close connection with where it all began as an online bookseller.
As well as still selling books directly, Amazon's marketplace is used by many third-party booksellers, and the company remains the leader in e-readers with its Kindle line launched back in 2007. That's why it's so harmful for the self-proclaimed "Earth's biggest bookstore" to have, apparently, just been caught destroying old and rare books to train AI – and even celebrating the fact with a fun cartoon logo design.
Amazon built its business on selling books, first coming online as a bookstore back in 1994. Now, according to an investigation by 404 Media, it's tearing old and rare books apart so it can feed them its AI models.
Journalists hid an Apple Airtag inside a book in a bulk shipment of 1,000 books ordered via Biblio, a marketplace where buyers can remain anonymous, after the seller said they suspected that the order was made by an AI company. The tag was then traced to Amazon's LAS8 warehouse in Las Vegas, reportedly the location of a scanning operation called VGT3.
404 reports cites workers at the facility, who say their work involves scanning books to provide training data for Amazon's AI. To speed up the process, they would cut the bindings off books.
In an online forum, a VGT3 employee said that “all we do is scan books,” and that “some are assigned to cut books, and others go to receive where they get books and scan the bar codes.”
That practice would be damaging enough for Amazon's brand image, but to make the optics even worse, the warehouse team reportedly has a logo design that depicts a Tyrannosaurus rex devouring a book. The tone-deaf image reinforces the irony of a bookseller now consuming books for profit, including rare and irreplaceable volumes.
A couple of years ago, a backlash led Apple to hastily withdraw an iPad Pro advert that showed a machine crushing physical objects like musical instruments, books and camera lenses. In Amazon's case, it seems the crushing machine really exists on an industrial scale as a core part of its AI development.
The revelation of such cultural vandalism raises more doubts about AI developers' claims that their technology is good for human creativity. It's also likely to alienate authors, bibliophiles, and customers by clashing with Amazon’s identity as a platform for readers.
The books bought via Biblio were not necessarily highly valuable in monetary terms, but they were defined as rare. "There are a lot of other types of value," the bookseller is quoted as saying. "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.”
Rare and old books not only offer "clean” pre-2022 text not polluted by machine-written content, but they can also contain data that rivals don't have, making them potentially valuable training material.
Courts have ruled that training AI on legally purchased books can qualify as fair use, but that doesn't change the fact that Amazon shredding rare works is highly damaging for its public image and brand even if technically legal.
Non-destructive scanning or licensing digital archives could achieve the same goal without destroying cultural artifacts. Rivals also destroy books to train AI, but Anthropic has claimed that it avoids using rare or antique books, which makes Amazon’s approach look particularly reckless.
“Amazon purchases books through commercial channels to help develop and improve the products and services our customers use,” an Amazon spokesperson said in a statement.
