Booksellers Suspect AI Firms Are Destroying Rare Books for Training Data

Ars Technica AI · rss · 2026-08-12

According to Ars Technica, booksellers suspect that AI companies are buying and destroying rare books on a large scale to rapidly acquire high-quality, long-form training text.

The article notes that the cheapest and fastest way to scan books involves cropping and scanning torn-out pages before discarding them. While the exact scale of this destructive scanning is unconfirmed, book lovers fear it is happening widely, leading to the permanent loss of physical copies.

Original post →

More from AGI Musings

AGI Musings channel →