An investigation by 404 Media, with assistance from Simon Willison, revealed that Amazon is acquiring large quantities of books, including rare ones, to scan them for AI training data. The process involves destructively scanning the books at an Amazon facility in Las Vegas, identified as VGT3 within the LAS8 complex. This practice aligns with the broader industry trend of AI companies seeking diverse training materials beyond readily available internet text, aiming to avoid issues like model collapse. AI
IMPACT Confirms destructive data acquisition practices by major tech firms, potentially impacting the value of physical media and the availability of diverse training data.
RANK_REASON Investigation reveals a major company's practice of acquiring and destroying books for AI training data.
Read on Mastodon — fosstodon.org →
- 404 Media
- AirTag
- Bibliothèque nationale de France
- Enciso
- Las Vegas
- Lawrence F. Katz
- Mastodon
- VGT3
- Anthropic
- Claude Fable-5
- Hugging Face
- LAS8
- OpenAI
- Qwen 3.8-27B
- Simon Willison
- AI Training Facility
- rare book
AI-generated summary · Google Gemini · from 15 sources. How we write summaries →