Judge approves Anthropic’s $1.5 billion copyright settlement

A federal judge has given final approval to Anthropic’s $1.5 billion settlement with authors and book publishers, Reuters reported Monday. For investors watching the AI sector, the deal removes one legal overhang for Anthropic but leaves the bigger copyright fight around model training unresolved.

Judge Araceli Martinez-Olguin of the U.S. District Court for the Northern District of California approved the settlement, according to Reuters. The case had previously been handled by Judge William Alsup, who granted preliminary approval last year before retiring.

The settlement covers an estimated 500,000 works and is set to pay $3,000 per work to authors and publishers with rights to those books. TechCrunch reported that the deal is believed to be the largest copyright settlement in U.S. history.

What the case was about

The lawsuit centered on books Anthropic used while building training data for its AI systems. Training data is the material used to teach an artificial intelligence model how to recognize patterns in language, images or other information.

Alsup ruled that Anthropic had unlawfully downloaded and stored millions of copyrighted books, according to TechCrunch. The court found a difference between two ways Anthropic built its book collection: buying and scanning books, and downloading books from pirate sites including Library Genesis and Pirate Library Mirror.

Alsup concluded that buying and scanning books was permitted, while the use of pirated copies created a separate legal problem. That piracy issue was allowed to move toward trial, according to TechCrunch, before Anthropic agreed to settle rather than face a jury’s decision on damages.

The fair-use ruling still matters

The most closely watched part of the case was not the payout. It was Alsup’s conclusion that using copyrighted text to train an AI model can qualify as fair use. Fair use is a legal doctrine that can allow limited use of copyrighted material without permission, depending on factors such as purpose, amount used and market effect.

That ruling was a meaningful win for Anthropic and the broader AI industry, according to TechCrunch. AI companies have argued that training models on large collections of text is a transformative use, meaning the material is used to create something different from the original books.

The settlement does not turn Alsup’s fair-use decision into nationwide law. Because the case ended before an appeal, the ruling remains a district court decision rather than binding precedent for other courts. Other judges can still reach different conclusions in cases involving different companies, datasets or facts.

Other AI copyright cases are still active

The approved settlement closes this case, but it does not settle the industry’s legal exposure. TechCrunch reported that copyright lawsuits are still pending against companies including Google, Meta, Midjourney and OpenAI over the use of copyrighted material in AI training.

Last week, publishers and authors including Hachette, Cengage, Elsevier, Scott Turow and S.C.R.I.B.E. filed a class action lawsuit against Google, according to TechCrunch. The lawsuit accuses Google of using copyrighted works to train Gemini, its AI platform.

For AI companies, the distinction matters: courts may treat the act of training a model differently from the way training material was obtained. For investors, that means legal risk in AI is unlikely to move in a straight line. One company can resolve a damages fight while the broader rules for the sector remain unsettled.

This story draws on original reporting from TechCrunch.