{
  "id": 1521065,
  "title": "Amazon, which started off selling books, is destroying rare texts to train AI",
  "url": "https://urgent.news/2026/08/17/amazon-which-started-off-selling-books-is-destroying-rare-texts-to",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-17T16:38:44.000Z",
  "source": {
    "name": "TechCrunch",
    "slug": "techcrunch",
    "url": "https://techcrunch.com/2026/08/17/amazon-once-an-online-bookseller-is-destroying-rare-books-to-train-ai-models/"
  },
  "original_language": "en",
  "account": "Amazon, initially a bookseller, is now acquiring rare texts to train artificial intelligence (AI), according to 404 Media. The company removes the spines of these books and scans them for AI development, as revealed by a tracking device placed within a rare book that eventually reached an Amazon facility in Las Vegas. The facility, identified as VGT3, bears a dinosaur holding a book in its claws.\n\nAmazon disclosed to 404 Media that they procure books via commercial channels to enhance their products and services. The company's demand for vast amounts of text is immense, essential for training their Large Language Models (LLMs). These AI models have already absorbed the majority of text available on the internet; however, they still require significantly more data to function optimally.\n\nRare books, particularly those out of print or otherwise inaccessible online, present a valuable source of training data for these AI systems. This is especially true since anything published before 2022 could potentially have been generated by an LLM. When LLMs train on AI-generated text, they risk \"model collapse,\" a phenomenon where the quality of an LLM's outputs deteriorates after ingesting too much AI-generated text.",
  "summary": "Rare books are incredibly valuable for training LLMs, since these models have already trained on whatever's available online.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}