A judge just gave the final stamp of approval on one of the biggest payouts in copyright history.

Anthropic, the AI company behind Claude, can now start paying out $1.5 billion to authors and publishers. This wraps up a legal fight that started over how the company built its AI.

Settlement

On Monday, a federal judge approved Anthropic’s massive settlement. This closes out a class action lawsuit that authors and book publishers filed against the company.

The case centered on whether Anthropic broke copyright law when it trained its AI models on books. 

Judge William Alsup handled the case early on. He gave preliminary approval to the settlement last year. But Alsup has since retired. 

So Judge Araceli Martinez-Olguin took over and signed off on the final deal this week. Now Anthropic has to start cutting checks.

Payouts

The settlement covers an estimated 500,000 works. Each one is worth $3,000.

That money goes to the authors and publishers who own the rights to those books. Many experts believe this is the largest copyright settlement in U.S. history.

Sounds like a win, but not everyone agrees.

Authors 

The legal outcome wasn’t actually a loss for Anthropic; it was mostly a win. Judge Alsup ruled on the core issue at the heart of the case. 

He decided that training an AI model on copyrighted books counts as fair use. That’s a legal term that basically means it’s allowed, even without permission from the original authors.

This ruling left many in shock. It was seen as a green light for tech companies to keep training their models on copyrighted material.

So if training on books is legal, why did Anthropic have to pay $1.5 billion?

Also read: Grammarly Faces Class Action Lawsuit from Writers

Data Collection

Anthropic built its book collection two different ways. First, it bought and scanned physical books. That part was fine and totally legal.

Second, it downloaded books from pirate websites. Sites like Library Genesis and Pirate Library Mirror are known for hosting stolen copyrighted content. 

Judge Alsup found that this method broke the law. That act of piracy was serious enough to go to trial. Rather than risk a jury deciding damages, Anthropic chose to settle.

In short, Anthropic lost the battle but won the war about AI training. 

Laws

This settlement only applies to Anthropic. It doesn’t create a binding rule for the whole AI industry.

Because Alsup’s ruling came from a single district court, and since Anthropic settled instead of going to trial, the case will never reach an appeals court. 

That means it can’t become official legal precedent. Other judges handling similar cases can still reach totally different conclusions. They’re not required to follow Alsup’s reasoning.

More Lawsuits 

Several other tech giants are facing similar legal battles right now. Companies like Google, Meta, Midjourney, and OpenAI are all dealing with lawsuits over their AI training practices.

Authors and publishers across the industry are pushing back hard. Just last week, a group of major publishers filed a new lawsuit against Google. 

The list includes big names like Hachette, Cengage, and Elsevier. Author Scott Turow joined the case too, along with a group called S.C.R.I.B.E.

Google allegedly used their copyrighted books to train its AI platform, Gemini, without permission.