Airtag Reveals Amazon Trashing Rare Books to Train AI

An AirTag exposed the truth. Amazon is trashing rare books, and the proof was tucked inside a single volume. A bookseller agreed to plant the tracking device in a rare book that was part of a bulk order, then watched where it went. The trail ended at an Amazon AI training facility in Las Vegas. There, a team was tearing books from their spines and scanning pages. It's a shocking discovery. But the device didn't lie, and the journey was over before anyone could intervene.

For roughly a year, booksellers have suspected that AI firms were buying huge lots of rare books, scanning them, and then destroying them to feed their models. But suspicion isn't proof. The Airtag changed that. The device was tracked to a warehouse known as VGT3, where workers openly discussed the work online. A logo on the door showed a Tyrannosaurus rex preparing to devour a book. The tone-deafness of that mascot, given the escalating backlash over destructive book scanning, was apparently lost on the team.

Amazon deflects while the scanner hums

Amazon declined to comment on the findings. The company's only statement was a generic line about purchasing books through commercial channels to develop and improve products and services. It did not mention AI training specifically. But the context is hard to ignore. Amazon is developing frontier AI models, and those models need massive amounts of unique training data to compete with Google, OpenAI, and Anthropic.

Right now, firms guard their training data fiercely. An edge can vanish if a rival trains on the same public text. But rare books offer something that scraped websites don't: original content that's difficult to find anywhere else, and that's a supply that's largely untapped by the biggest players in AI. Anthropic and xAI have both stated publicly that they are not training on rare or antique books. So that leaves room for a competitor like Amazon to quietly scoop up the supply. It's a narrow window.

The shortage that almost shut it down

Online forum posts from VGT3 workers suggest the operation nearly ran dry earlier this year. Amazon had run low on books to scan, and the anxiety among staff was palpable as they watched the dwindling inventory and wondered if the warehouse might shut down if more books couldn't be found. It was a close call. At one point, the supply completely ran out. Yet the facility is still operational, and bulk orders of rare books delivered there are systematically destroyed, so the machines keep humming and the workers keep turning pages into pixels. Strange, isn't it? But that's the reality.

The irony is thick. Booksellers carefully sourced these works, assessing their value, and preserving them. Now a crew in Las Vegas is tearing them apart for training data. Some Amazon workers described the gig as a "nice" opportunity for those drawn to a humdrum job with flexible hours. Book lovers see it differently. One staunch critic, Michael Burry, reportedly called the practice "evil incarnate."

Why ISBNs are the new gold

The investigation also helped firm up a long-standing bookseller theory about how AI firms choose which books to buy. Booksellers had suspected that AI companies were targeting books with ISBN numbers to ensure the highest volume of unique works in their training sets, and that suspicion now holds more weight. But the Airtag shows Amazon destroying rare books. That's one thing. Online discussions from VGT3 workers added an important detail, though: they were trained to scan barcodes or ISBNs before scanning the pages, and that simple step reveals a cold, mechanical logic behind the whole process. It's not random. It's not about quality. It's about quantity, pure and simple.

That practice gives further credence to the theory that AI companies are methodically working through the list of ISBNs to scan every printed book in the world. It's a checklist approach. Quantity matters more than quality.

The books being bought aren't necessarily prized first editions of celebrated works. Instead, AI firms often target older books with lower monetary value, such as works never translated from a language that's not widely used today, or books that were never popular enough to be widely distributed.

But these works still carry weight. The bookseller who planted the Airtag noted that these books may hold "historical value, intellectual value, sentimental value" that AI firms overlook, and that's a point worth sitting with. A rare book's value can derive from "all sorts of things" that "the AI companies don't care about," so they're missing the point entirely. They just want the content as a bunch of words strung together. That's it.

Reddit splits over the ethics

On Reddit, the debate got messy. Some argued that destroying rare books isn't a big deal, especially since some have sat on shelves for decades gathering dust. "You were not going to buy that old paper book," one commenter said. But others pushed back hard. One poster lamented that "an obscure book from 1700 is now a museum piece and may reveal day-to-day stuff that we didn't know." The real problem, they argued, is that tiny details and major historical insights will be lost to AI greed. AI firms will "never share the contents" of what they scan, the poster said, because they don't want anyone else training on the same works.

One commenter warned that AI models would only serve up "warped, censored, and paywalled fragments of ideas" from forgotten works. But another user admitted the whole notion of an AI literally eating books to grow more powerful is "pretty dystopian," and that admission lands with a thud because it's not a tech critic saying it, it's someone inside the conversation, wrestling with their own discomfort. So the back-and-forth captures a growing unease about what we're sacrificing in the race for smarter models. That unease is real. We can't shrug it off.

"It would be a much bigger problem if a book like that were bought for destruction, having survived this long.""

Scottish bookseller Derek Walker told the BBC that AI firms should distinguish between works that won't be missed much and lesser-known antique works that may be "the only known surviving example of an edition." His point cuts to the heart of the issue. Not every text needs to be saved. But some books have survived centuries, wars, and neglect. Watching them get shredded for training data, after they've lasted this long, is a different kind of loss.

The transparency problem

For AI firms guarding their training data, there's little incentive to discuss bulk buying with booksellers. So they hide. Many use services that mask their identities as buyers, making it hard for sellers to know who's on the other end, and that's a deliberate wall of secrecy. Allowing booksellers to weigh in on which works get destroyed would require a level of transparency that may seem riskier than a possible reputation hit, but it's hard to imagine a more direct route to trust.

But the Airtag reveals Amazon trashing rare books in a way that's hard to dismiss. It's blunt. The tracking device turned a suspicion into evidence, showing that a major tech company is willing to destroy physical books, which are finite and irreplaceable, to gain an edge in AI, and that's a fact we can't just wave away. The booksellers who supplied those books may have made good money on the deals. So was that cash worth the loss? That question will linger.

The ethical dilemma won't disappear. It's a stubborn stain. Booksellers know how to evaluate many types of rare works to determine their worth, drawing on decades of expertise and careful provenance checks, yet AI firms seem to be skipping that step entirely, hunting low-cost, unique ISBNs to complete their checklists. So the result is a system where historical artifacts are treated as raw material. And the only proof we have of it, for now, is a small tracking device tucked inside a single book,but we can't use a dash, so we'll say: a small tracking device tucked inside a single book.

Frequently Asked Questions

What did the AirTag reveal about Amazon's treatment of rare books?

The AirTag revealed that Amazon is trashing rare books to train AI. It was tracked to an Amazon AI training facility in Las Vegas, where workers were tearing books from their spines and scanning pages.

Why are rare books valuable for AI training according to the article?

Rare books offer original content that is difficult to find elsewhere, unlike scraped websites. They provide unique training data that competitors may not have, giving Amazon an edge in developing frontier AI models.

How did Amazon respond to the findings about the rare book destruction?

Amazon declined to comment on the findings. Their only statement was a generic line about purchasing books through commercial channels to develop and improve products and services, without mentioning AI training specifically.

What detail from VGT3 workers supports the ISBN theory?

VGT3 workers were trained to scan barcodes or ISBNs before scanning pages. This mechanical step indicates AI firms are methodically working through ISBN lists to scan every printed book, focusing on quantity over quality.

What ethical concern did Reddit users raise about destroying rare books?

Reddit users argued that obscure books may hold historical value and reveal unknown daily details, which would be lost. They also noted AI firms will never share the contents, so the information becomes inaccessible, leading to warped, censored, and paywalled fragments of ideas.