AI companies are reportedly shredding millions of books after using them to train AI models. This process involves acquiring large quantities of physical books, scanning their contents for training data, and then destroying them to prevent the information from being accessed by competitors or the public.

The practice has been facilitated by middlemen who purchase books in bulk from libraries, bookstores, and online retailers. These intermediaries then sell the books to AI firms, which use them to train their models before disposing of them. This method allows companies to keep their data acquisition strategies confidential.

A recent investigative report from 404 Media reveals that these companies are reportedly purchasing books in large volumes, often from sources that are not aware of the ultimate use of the materials. The report highlights that this practice has been ongoing for several years and has raised questions about the sustainability and ethical implications of such methods.

The consequences of this practice include increased costs for AI firms, potential vendor lock-in with book suppliers, and challenges in governance and transparency. Market reactions have been mixed, with some stakeholders expressing concern over the environmental impact of shredding large quantities of books.

The situation remains under active discussion, with no clear regulatory framework in place to govern the acquisition and disposal of books for AI training. As the demand for AI models continues to grow, the industry may need to address these issues more directly in the coming years.