Amazon Cuts Spines Off Rare Books to Train AI
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Amazon is buying rare books, cutting off their spines, and scanning them for AI training, according to 404 Media.
- 404 Media placed a tracking device inside a rare book, which arrived at an Amazon facility in Las Vegas called VGT3, identifiable by a dinosaur holding a book in its claws.
- Amazon told 404 Media it "purchases books through commercial channels to improve the products and services customers use."
- LLMs need unfathomably large amounts of text and have already exhausted what they can scrape from the internet, with Anthropic specifically noted as having pirated books.
- Rare books, especially out-of-print titles absent from the web, are prized training data because any text published before 2022 was guaranteed to be human-written, avoiding "model collapse" from AI-generated input.
Why it matters: Amazon is physically destroying irreplaceable texts — books that don't exist online and may not exist elsewhere — to feed LLMs already running into data scarcity. The practice adds a tangible cultural-heritage cost to the AI training arms race and the model-collapse problem that arises when models run out of fresh human-authored data.
Ask SkimNews
