AI firms buy and shred rare books

Several AI start‑ups have quietly bought collections of out‑of‑print works, scanned them, and then shredded the physical copies to avoid copyright entanglements.

The firms argue that removing the originals eliminates legal risk while providing high‑quality data for large‑scale language models.

Heritage groups warn that the loss of rare volumes erodes the public record and could set a dangerous precedent for cultural preservation.

The controversy has sparked calls for clearer regulations on data sourcing for AI training.