A recent investigation led by Emanuel Maiberg of 404 Media has exposed a systematic process in which thousands, and potentially millions, of books are being destroyed to feed the growing demand for AI training data. The report details a complex, globalized supply chain that operates with significant opacity. Books are often acquired through anonymous online marketplaces before being funneled into large-scale scanning facilities.

At these facilities, the physical integrity of the books is compromised; bindings are removed so that individual pages can be fed into high-speed scanners. This process effectively turns the books into digital data sets, after which the physical copies are discarded. The investigation characterizes this activity not as a result of malicious intent, but rather as a byproduct of institutional indifference toward the cultural and physical value of the printed works.

While the report highlights the scale of the operation, it notes that the supply chain is designed to be difficult to trace, making it challenging to identify the specific entities ultimately responsible for the final AI training applications. The findings underscore a growing tension between the rapid development of artificial intelligence and the preservation of physical media, as the industry continues to prioritize the acquisition of massive amounts of text for machine learning purposes.