The rapid development of artificial intelligence has created a massive demand for high-quality training data, leading tech companies to digitize millions of physical books. According to an investigation by France 2, this process frequently involves the destruction of the original volumes once they have been scanned. While the digitization allows for the creation of vast datasets necessary for training AI models, the physical disposal of these books has sparked a debate about the long-term impact on cultural heritage.

Critics of the practice point to the potential loss of rare, out-of-print, or second-hand books that may not exist in any other format. By prioritizing the digital extraction of text, the physical artifacts are discarded, effectively removing them from circulation permanently. The investigation highlights a tension between the technological advancement of AI and the preservation of physical media. While proponents of AI development argue that digitizing these works makes knowledge more accessible and searchable, the report emphasizes that the destruction of the physical copies poses a unique risk to historical and literary archives. Currently, there is no standardized industry practice for preserving the physical books after they have been processed, leaving the future of these collections uncertain.