Hidden Airtag reveals Amazon is trashing rare books to train AI (arstechnica.com)

A recent investigation by 404 Media has revealed that Amazon may be purchasing and subsequently destroying rare books to train its artificial intelligence models. This came to light after an Airtag was discreetly placed in a rare book, which led to its tracking at an Amazon AI training facility in Las Vegas. The facility, where teams were observed dismantling books for scanning, raises ethical concerns about the destruction of valuable literary works in the pursuit of AI advancement. An emblem at the facility humorously depicted a dinosaur about to consume a book, highlighting the tone-deafness amidst the ongoing backlash against such practices. The significance of this discovery lies in the implications for the AI/ML community, where access to unique and diverse training data is crucial for developing competitive models. As companies like Amazon strive to keep pace with rivals such as Google and OpenAI, the acquisition of rare texts could provide a substantial edge. However, this aggressive data sourcing strategy also draws scrutiny against the backdrop of ethical AI development, especially as competitors like Anthropic and xAI publicly refuse to exploit rare or antique books for training purposes. The situation raises critical questions about the balance between innovation and preservation in the evolving landscape of AI.