An investigation by 404 Media has uncovered Amazon's practice of acquiring numerous rare books, subsequently removing their spines, and scanning their contents for the explicit purpose of training artificial intelligence. This discovery was made possible through a tracking device discreetly placed within a rare book, which ultimately led to an Amazon facility located in Las Vegas.
The facility, identified as VGT3, notably features a logo depicting a dinosaur holding a book in its claws. When approached for comment, Amazon provided a statement to 404 Media, asserting that it "purchases books through commercial channels to improve the products and services customers use."
Companies like Amazon require an unparalleled volume of textual data to effectively train their large language models (LLMs). As the vast majority of accessible internet content has already been ingested—and in some instances, such as with Anthropic, illegally pirated books have been utilized—rare books, especially those that are out of print or not readily available online, present a highly desirable new source of training data.
These historical texts are particularly valuable due to the certainty that any material published before 2022 was not generated by an LLM. This distinction is vital because when LLMs are trained on AI-generated text, they face the risk of "model collapse," a phenomenon characterized by a significant degradation in the quality of an LLM's outputs after prolonged exposure to and ingestion of too much synthetic text.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.