Anthropic agrees to landmark $1.5B copyright settlement in IP infringement lawsuit

Anthropic agreed to pay $1.5 billion to settle a major copyright lawsuit after authors said the company used pirated books to train its Claude AI chatbot without permission.

The proposed class-action deal is the biggest copyright payout ever reported and the first major settlement tied to artificial intelligence. If Judge William Alsup in San Francisco approves it, the deal will pay hundreds of thousands of authors and set a new standard for how AI companies use copyrighted material.

This comes after a U.S. judge allowed a group of authors to proceed with a lawsuit against AI firm Anthropic, alleging their books were used without consent to train its artificial intelligence model. The authors, all established professionals, claim Anthropic used their copyrighted works to train its Claude chatbot without permission or compensation.

They argue the company went too far by leveraging their stories to make the AI sound more human, even imitating their writing styles and ideas.

Authors secure record payout in copyright case

The $1.5 billion settlement covers about 500,000 pirated books the company downloaded from online shadow libraries like Library Genesis, Pirate Library Mirror, and the Books3 dataset. These online libraries are popular sources of unauthorized digital copies of books that researchers and AI developers use for their datasets. For a long time, writers, publishers, and advocacy groups have criticized the repositories for undermining copyright protections.

The deal says that each writer whose book Anthropic illegally used as training data for its AI will receive about $3,000 in compensation. If further investigation discovers any additional work the company used in its dataset, the total amount could skyrocket beyond the initial $1.5 billion.

Anthropic’s case began in 2024 when authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson sued the company for copying their books without permission. Later on, hundreds of thousands of other writers filed a lawsuit, claiming the company used their works to train its flagship chatbot, Claude. 

The authors flagged Anthropic’s ignorance of licensing the material legitimately despite its financial status, as seen in its powerful investors, like Amazon and Google’s parent company Alphabet.

Judge William Alsup ruled in June that using copyrighted material to train AI models could be considered fair use if the works were transformative. But he also said that Anthropic violated copyright law when it downloaded and stored over 7,000,000 pirated books in a central library.

The ruling exposed the company to damages if the lawsuit went to trial as scheduled in December because legal experts warned that Anthropic could have to pay hundreds of billions of dollars.

Anthropic avoids trial but still faces legal risks

The terms of the $1.5 billion settlement require Anthropic to destroy every copy of the pirated books to prevent it from continuing to benefit from unauthorized material. However, experts say the agreement doesn’t protect the company from fresh lawsuits. Authors could still sue if the Claude system generated outputs that reproduce copyrighted passages or infringe on their works. 

Anthropic said the deal did not include an admission of liability, but rather that it wanted to move forward without the uncertainty of a lengthy trial. The company’s deputy general counsel, Aparna Sridhar, explained that Anthropic is still committed to producing safe, trustworthy, and supportive of human progress. She said the company wants to concentrate on developing tools that improve science and serve the public rather than spending years locked in court battles.

Other AI companies like OpenAI, Microsoft, and Meta still face lawsuits that raise nearly identical copyright concerns. A judge in San Francisco recently allowed a lawsuit against Meta to continue, saying the use of copyrighted material without permission is unlawful “in many circumstances.”

Settling saved Anthropic the possibility of dragging the dispute to the U.S. Supreme Court because this could’ve brought massive changes to national laws around how fair use applies to artificial intelligence.

The smartest crypto minds already read our newsletter. Want in? Join them.