Can AI be trained on copyrighted material without permission? Yaniv and Chris break down two explosive legal battles shaking up the AI industry.
First, Aardman Animations (of Wallace & Gromit fame) and the BBC threaten legal action over AI scraping, while in the US, Anthropic is facing a lawsuit for using pirated books to train its Claude models.
While the court ruled that training AI on legally acquired content is fair use, even if the model “remembers” what it saw, the real issue is that Anthropic didn’t pay for the books.
In this clip:
* What the U.S. court actually said about training AI on copyrighted books
* Why Anthropic could face up to $750 billion in damages but still win the bigger war
* If AI is “remembering” content the same as human learning?
* What this ruling means for founders training their own models
* The thin legal line between “fair use” and “piracy” in the AI age
Clip from Reacts episode 'Anthropic's Huge AI Copyright Win + Speech to Text Better Than Ever, Peter Thiel's Tech Dystopia': https://youtu.be/Tb7CbvrZmRw
🎙 Want more insights like this?
Subscribe to The TSP Mailing List for top episodes, actionable startup tips, and exclusive resources straight to your inbox! Don't miss out. Subscribe here. https://thestartuppodcast.beehiiv.com/subscribe
#FairUse #AICopyright #Anthropic #AITraining #StartupLaw #CopyrightLaw #Aardman #PerplexityAI #AIethics #AIstartup
