Judge signs off on $1.5B Anthropic deal, but fair use fight rages on
Curated by the Inblix editorial team
A federal judge gave final approval Monday to a $1.5 billion settlement between Anthropic and a class of authors and publishers, officially closing a landmark copyright case that was heading toward a messy trial over how the AI lab built its training library. Judge Araceli Martinez-Olguin signed off on the deal after her predecessor, Judge William Alsup, ruled last year that Anthropic had illegally downloaded millions of copyrighted books from pirate sites like Library Genesis—even as he separately found that training an AI model on copyrighted text constitutes fair use.
The payout structure is straightforward: $3,000 per work across roughly 500,000 works. It’s being called the largest settlement in U.S. copyright history, but many creators aren’t celebrating. The money resolves the piracy question without the company having to admit broader liability, and the settlement preempted a trial where a jury might have awarded far heftier damages. Anthropic had two buckets of training data—books it purchased and scanned, and books it grabbed from shadow libraries. The first was legal; the second, Alsup said, was not.
The settlement check clears, but the legal precedent is basically nonexistent. Because Anthropic settled, Alsup’s fair use ruling will never reach an appeals court, so it doesn’t bind anyone else. Other judges are free to look at the same question and come to completely different conclusions. That means OpenAI, Meta, Midjourney, and Google are still very much in the hot seat—with no real clarity on whether scraping copyrighted material to train generative models is a protected activity or industrial-scale infringement.
Google, for its part, got dragged into the fight just last week. Hachette, Cengage, Elsevier, author Scott Turow, and writing collective S.C.R.I.B.E. filed a class action accusing the company of using their copyrighted works to train Gemini. The timing underscores how little the Anthropic deal actually resolved for the industry. One door closes, but half a dozen others are still wide open—and no one knows yet which way the next judge will jump.
💡 Key Takeaways
- The $1.5 billion settlement pays $3,000 per work across 500,000 works but leaves open the fundamental question of whether training AI on copyrighted text is fair use.
- Judge Alsup ruled that using pirated books for training was illegal on its face, while separately finding that training itself qualifies as fair use—a split that let Anthropic settle and avoid a jury.
- Because the settlement prevents an appeal, Alsup’s fair use reasoning won’t become binding precedent, leaving Google, OpenAI, Meta, and others to fight identical battles with no clear legal guidepost.
- The Google lawsuit filed last week by Hachette, Scott Turow, and others shows publishers aren’t backing down and are willing to sue every major AI developer one by one until a definitive ruling lands.
Keep reading: See related articles below for more coverage on this topic.
Get smarter about AI
The sharpest AI news, curated daily. Delivered free to your inbox.