AI Companies Destroy Physical Books – Scan Rare Books Before It's Too Late

A deeply troubling trend has emerged within the AI industry: companies are purchasing large quantities of physical books solely to use them as training data for their large language models, and subsequently destroying them. What makes this especially alarming is that many of these books are rare or one-of-a-kind volumes representing irreplaceable cultural heritage.

Anna's Archive, a well-known digital library preservation project, has published an urgent call to action for the public and the tech community. They argue that we are at a critical juncture where the window to preserve these works is rapidly closing. The project is urging volunteers, libraries, and organizations to prioritize the digitization of rare books before they end up in the hands of AI companies or are otherwise lost.

The issue is complex and multifaceted. On one hand, AI companies have a genuine need for high-quality textual data to train their models effectively. On the other hand, the physical destruction of books — particularly rare ones that may never have been digitized — risks permanently erasing works that exist in only a handful of copies worldwide. Critics argue that there is a profound ethical problem with destroying cultural heritage for commercial gain, regardless of the technological benefits.

The discussion on Hacker News, garnering over 179 comments and 258 upvotes, demonstrates how widely this issue resonates across the tech community. Many users are calling for regulations and greater transparency regarding how AI companies source and handle physical materials. Others are emphasizing the importance of supporting preservation initiatives like Anna's Archive to ensure that humanity's collective knowledge is safeguarded for future generations.