4 hours ago
- AI companies are buying rare books, scanning them with high-speed machines that cut off spines, and then shredding the originals.
- A service called ISBNdb facilitates bulk orders of up to a million books and keeps buyers anonymous.
- Pre-2022 books are considered premium because they contain no AI-generated text.
- A federal judge ruled the practice as fair use, reasoning that eliminating the original means only one copy exists at a time.
- Anthropic hired the former head of Google Books partnerships to obtain 'all the books in the world.'
- Rare books with few surviving copies—those that survived wars, fires, and centuries—are being destroyed for AI training data.
- ISBNdb acknowledges the negative headline potential but built its business around discreet operations, including NDAs.
- The author argues this practice is worse than previous AI data scraping because the destruction of physical originals is irreversible.
- The legal ruling is expected to accelerate the destruction of rare books for AI training.