Amazon Destroys Rare Books to Train AI, Report Finds

A recent investigation by 404 Media has revealed that Amazon is buying rare books, removing their spines, and digitizing them to improve its AI models. The research uncovered that a tracking device in one of these rare books pointed back to an Amazon site in Las Vegas, called VGT3, which has a dinosaur logo holding a book. Amazon confirmed this practice, explaining that it 'acquires books through commercial sources to enhance customer offerings.' The push for these rare texts stems from the need for large datasets to train language models that have already utilized a lot of internet content, sometimes even illegally. Since rare and out-of-print books are less likely to have AI-generated content, they pose a risk of ethical issues related to preserving cultural artifacts.