Amazon reportedly destroying rare books to use as AI training data

Amazon, originally founded as an online bookseller, is now reportedly destroying rare books to use as training data for large language models. Rare and out-of-print books are considered highly valuable for AI training because much of their content is not available on the internet. LLMs benefit from such offline material as it expands the breadth and diversity of their training datasets. The move has drawn attention given Amazon's origins as a bookseller and the irreplaceable nature of rare literary works.