
No Priors AI The Legal Price of Progress
Nov 9, 2025
A $1.5 billion settlement sparks a debate on the legal costs of AI innovation. Discussions reveal Anthropic’s controversial methods, using both purchased and pirated books for training. The fair use ruling highlights the distinction between legal and illegal sources. There’s a look at the implications for future lawsuits against AI firms, while exploring practical challenges in compensating creators. Despite these complexities, there’s a call to embrace AI's benefits responsibly, navigating the evolving landscape of copyright and technology.
AI Snips
Chapters
Transcript
Episode notes
Historic Copyright Settlement Changes The Game
- Anthropic agreed to a historic $1.5 billion settlement over copyright claims tied to pirated book libraries.
- The case forces the industry to confront legal costs of using copyrighted training data and sets a major precedent.
How Claude Got Its Writing Tone
- Jaeden explains Anthropic scraped pirated "shadow libraries" of millions of books to rapidly improve Claude's writing tone.
- They then also bought books and used robots to scan and transcribe pages to cover their tracks.
Fair Use Ruling For Transformative Training
- The judge ruled training on copyrighted material can be fair use when the model transforms knowledge rather than reproducing works.
- That decision separates legality of purchased material ingestion from the illegality of using pirated sources.
