In a landmark decision that could reshape how AI companies source their training data, Anthropic agreed to a $1.5B settlement with authors who accused the startup of using pirated books to train its Claude models.
Why it matters
Expect a stronger shift to licensing over scraping, renewed interest in synthetic data, and ripple effects across other lawsuits.
What to watch
How courts treat model weights, dataset destruction, and whether smaller labs can comply without chilling competition.