Anthropic to Pay $1.5 Billion to Authors for AI Training Data
AI company Anthropic plans to distribute $1.5 billion to authors whose works were used to train its language models, including several prominent Swiss writers. The compensation addresses the practice of acquiring used books, scanning them, and feeding the text into AI systems without prior permission. This move highlights ongoing tensions between the tech industry and creators over intellectual property rights in the age of generative AI.
Key facts
- Anthropic will pay $1.5 billion to authors
- Swiss writers are among those on the compensation list
- The payment is for using books to train AI language models
- Books were obtained through purchase of used copies
- The practice involves scanning and digitizing physical books
- No prior permission was obtained from authors
- This reflects broader disputes over AI training data and copyright
- Anthropic is an AI company based in the US
Entities
Institutions
- Anthropic
Locations
- Switzerland
- United States