Companies Destroying Millions of Books for AI Training Sparks Controversy
AI companies are acquiring millions of physical books, destroying them to scan their contents into training datasets for chatbots. This practice has raised ethical and legal concerns as it reshapes the used-book market and fuels copyright lawsuits. Intermediaries help these firms source large quantities of books while maintaining confidentiality due to the sensitivity around data acquisition methods. The surge in demand is driven by a race to preserve human-authored content before AI-generated text becomes more prevalent, with pre-2023 publications being especially valuable for training high-quality models. While some booksellers have benefited financially from increased sales, concerns persist about rare and out-of-print titles being permanently lost after scanning. Legal precedents suggest that destroying physical copies during digitization may be considered fair use under certain conditions, but recent settlements indicate the risks of using pirated content for training AI.
Topics: Artificial Intelligence Copyright Issues Used Book Market