Historia
agosto 17, 2026
La caza de datos de IA de Amazon pone en riesgo libros raros
Un rastreador oculto en un libro raro llevó a una instalación de Amazon donde, supuestamente, los trabajadores destruyen volúmenes para un escaneo rápido. Amazon afirma que las compras mejoran los productos para los clientes, pero los libreros temen que obras irreemplazables estén siendo sacrificadas por datos para IA.
The race for AI’s next trove of training text has collided with a far older scarcity: books that may survive in only a handful of physical copies.
For months, rare-book sellers had noticed unusually large, indiscriminate orders—often for obscure titles rather than prized first editions—and suspected AI companies were behind them. Their concern was not merely that books were being digitized, but that rapid scanning could turn one-of-a-kind cultural artifacts into waste. Some sellers argue the value of such works lies well beyond resale price: in forgotten languages, local histories and editions that may never have been duplicated.1
There was another path. The Internet Archive has long used careful, page-by-page handling for brittle and special-collection volumes, saying automated systems were unsuitable for them and that “clean, dry human hands are the best way to turn pages.”2 That approach is slower and more expensive—the very trade-off critics say a data-hungry AI industry is resisting.
Then came the tracker. 404 Media placed an Airtag in a shipment of rare books; it arrived at Amazon’s VGT3 facility in Las Vegas. Employees at the site said they receive large shipments, cut off bindings to speed scanning and destroy the printed books in the process.3 Reports say workers scan barcodes or ISBNs first, reinforcing booksellers’ suspicion that buyers are systematically seeking unique texts rather than carefully selecting books for preservation.
Amazon has not specifically said the books are used to train AI. Its response was broader: it “purchases books through commercial channels to improve the products and services customers use.”4 That leaves a sharp divide. Amazon’s formulation frames the buying as ordinary product development; preservation advocates see an opaque supply chain in which physical copies disappear while the resulting digital material remains closed off.
The dispute is not over whether old books can be digitized. It is over who decides which ones can be destroyed—and whether the speed of the AI race is becoming the deciding criterion.