AI companies are purchasing large quantities of rare books, scanning them with high-speed machines that cut off the spines, and then shredding the original copies to create training data. This practice, facilitated by a service called ISBNdb, allows orders of up to one million books while keeping buyers anonymous. A federal judge recently ruled this process as fair use, despite the destruction of irreplaceable originals, according to xcancel.com.
ISBNdb, which markets itself as a digital preservation service, offers non-disclosure agreements to clients and advises them to frame the destruction as preservation. Anthropic, an AI firm, hired the former head of Google Books partnerships to acquire extensive book collections. A bookseller told 404 Media that rare books with very few surviving copies, some that have endured wars and centuries, are being destroyed to improve AI-generated content such as marketing emails, per xcancel.com.
This development raises concerns about the irreversible loss of cultural heritage, as unlike digital content, rare physical books cannot be replaced once shredded. While AI companies have previously scraped internet data and music, the destruction of unique physical books marks a new level of impact. ISBNdb’s website acknowledges the controversy but continues to support the practice, highlighting the legal backing from the federal court ruling, according to xcancel.com.
ISBNdb’s business model enables the destruction of millions of rare books, with the federal court decision effectively allowing only one copy to exist at a time. This legal precedent is expected to accelerate the practice, putting countless rare and historical books at risk of permanent loss.