404 Media says it tracked a shipment of rare books to an Amazon warehouse in Las Vegas, where workers described a book-scanning operation tied to Amazon’s product and AI work. The investigation, by Emanuel Maiberg, is the only substantive source in this cluster; Techmeme and Hacker News surfaced the same report rather than independently confirming it. According to 404 Media, Amazon is buying large quantities of printed books, scanning them for AI training data, and destroying the physical copies in the process. The report says employees at the Las Vegas site described receiving large book shipments, cutting off the bindings so the pages can be scanned faster, and leaving the printed books unusable afterward. Amazon did not deny buying books. In a statement to 404 Media, an Amazon spokesperson said the company purchases books through commercial channels to help develop and improve products and services used by customers. The statement, as reported, does not specify the scale of the buying program, which Amazon products are involved, or exactly how the scanned material is used. The reporting method was unusually direct. 404 Media says it worked with a bookseller who had received a large Biblio order of around 1,000 books and placed an Apple AirTag inside one of the books before shipment. The seller was granted anonymity because they feared business harm from being identified. Biblio did not respond to 404 Media’s request for comment, according to the report. The tracked shipment, 404 Media says, moved through an airport in California and was later observed near Milwaukee before ultimately ending at Amazon’s Las Vegas warehouse. The report identifies the Amazon team at the location as VGT3 and says its logo includes a dinosaur holding a book. The material point is not the route itself, but that the tracked order allegedly connected an anonymous marketplace bulk purchase to an Amazon facility handling book intake and scanning. 404 Media frames the finding against its earlier July reporting on booksellers who said they had seen a historical spike in sales over the past year. Those sellers suspected AI companies were buying books in bulk because the orders were large, the titles appeared varied or random, and the buyers did not seem price-sensitive. Marketplaces can obscure the end buyer, leaving sellers unable to know who is ultimately acquiring the books. The AI rationale is straightforward, as 404 Media describes it: printed books may contain text that is not easily available online, and older books printed before 2022 are less likely to contain AI-generated text. That matters because model developers have sought large bodies of human-written material, while also facing concerns that training on synthetic output can degrade future systems through a recursive effect often called model collapse. The evidence in this cluster is still developing. 404 Media has an on-the-record Amazon statement acknowledging commercial book purchases, but the most sensitive details—the scale of the operation, the scanning workflow, and the destruction of books—come from 404’s investigation and employee accounts as summarized in the report. There is no second outlet here independently verifying the Las Vegas operation. Who benefits: Amazon’s AI and product teams would benefit if commercial book purchases provide access to structured, human-written text not readily available online. Booksellers may also see increased sales from large bulk orders, though the report says sellers often do not know the end buyer. Who's exposed: Amazon faces scrutiny over the destruction of physical books and the opacity of its acquisition pipeline. Marketplaces such as Biblio could also face questions from sellers if bulk orders are being routed into scanning operations without clear buyer disclosure.