A US court has approved a $1.5 billion settlement in a landmark copyright infringement lawsuit against Anthropic, an AI company accused of using pirated books to train its Claude models.
Background of the Case
The lawsuit began in 2024 when authors Andrea Bartz, Kirk Wallace Johnson, and Charles Graeber filed a class action suit claiming Anthropic used their books without permission. While the court ruled that training AI on legally obtained books falls under fair use, it found that using pirated books was unlawful.
Details of the Settlement
- Anthropic’s internal project, known as Project Panama, involved purchasing and scanning physical books for AI training.
- However, the company also sourced books from piracy sites such as Library Genesis and Pirate Library Mirror.
- The court ordered Anthropic to compensate affected authors and publishers about $3,000 per illegally used book.
- Legal fees were capped at under 7% of the settlement, significantly less than the requested 20%.
- The lead plaintiffs received reduced individual awards of $15,000 each.
Ongoing Legal and Industry Challenges
Despite this settlement, Anthropic faces additional lawsuits, including one from Reddit over unauthorized content scraping, and another from Chicken Soup for the Soul publisher alleging unauthorized use of copyrighted books by multiple AI firms.
Anthropic has also accused Alibaba of misusing Claude outputs to train its own AI models, and concerns have been raised about user tracking within Anthropic’s Claude Code.
Industry Context
Once considered a top AI innovator, Anthropic has recently restricted access to some models due to US export controls and faces rising competition from OpenAI, Google, and Moonshot AI, all advancing their own AI technologies.
