A hidden Airtag in a rare book revealed Amazon is systematically destroying books to train AI models, according to 404 Media. The investigative group connected with a bookseller who planted the tracking device in a rare book that was part of a bulk order. That book was later tracked to an Amazon AI training facility in Las Vegas, where workers were observed tearing books from their spines and scanning pages. The facility, labeled VGT3, had a logo depicting a Tyrannosaurus rex devouring a book, according to 404 Media. Amazon declined to comment on the findings, offering a generic statement about purchasing books to improve customer products. However, the company is developing frontier AI models that require vast amounts of unique training data to remain competitive with firms like Google, OpenAI, and Anthropic. Amazon workers reportedly faced a shortage of books to scan earlier this year, with some fearing the facility might shut down if supplies ran out. The facility is still operational, and bulk orders of rare books delivered there are systematically destroyed by the crew. Many book lovers are horrified by the practice, with some critics calling it 'evil incarnate.' Amazon workers, however, described the gig as a 'nice' opportunity for those seeking flexible hours. The practice has sparked debate over the ethics of destroying rare books for AI training. 404 Media suggested its investigation reinforced a theory among booksellers that AI firms are targeting books with specific ISBN numbers to ensure a wide range of unique works are included in training data sets. The group reported that Amazon workers were trained to scan barcodes or ISBNs before scanning books, giving further credence to the theory that AI companies are trying to systematically scan every printed book in the world by working through the list of ISBNs. For booksellers, the money may be good, but the risk that their carefully sourced collections will be used for destructive book scanning raises an ethical dilemma. They know how to assess the value of rare books, but AI firms seem to be skipping that step in hunting for low-cost, unique ISBNs to complete their checklists. The books that AI firms are apparently buying up aren’t necessarily the kind of prized first editions of celebrated works that are typically valued highly. Instead, they often target older books with lower monetary value, such as books that were never translated from a foreign language that’s not widely used today or books that were never popular enough to be widely distributed. These works may still have 'historical value, intellectual value, sentimental value' that AI firms overlooked, the bookseller who planted the Airtag told 404 Media. A rare book’s value can be derived from 'all sorts of things' that 'the AI companies don’t care about. They just want the content as a bunch of words strung together.'
Source: arstechnica
