Amazon has reportedly been purchasing large quantities of rare books, removing their spines, and scanning them to train AI models, according to 404 Media. The outlet placed a tracking device inside a rare book and traced it to an Amazon facility in Las Vegas. That facility, called VGT3, is marked with a dinosaur symbol clutching a book in its claws. In response, Amazon told 404 Media that it “purchases books through commercial channels to improve the products and services customers use.”

Training large language models requires an enormous volume of text, and companies like Amazon have already consumed much of what’s freely available online—Anthropic, for instance, has been accused of using pirated books. Rare books, particularly those that are out of print or absent from the internet, provide a fresh source of highly sought-after training material. These texts are especially appealing because anything published before 2022 is guaranteed not to have been written by an AI. When LLMs train on AI-generated content, they run the risk of “model collapse,” a phenomenon where the quality of their outputs deteriorates after exposure to too much synthetic text.