Court Rules Anthropic Did Not Breach Copyright in AI Training

A US judge has ruled that a tech company’s use of books to train its artificial intelligence (AI) system, without the explicit permission of the authors, did not breach copyright law. This landmark decision by a federal judge in San Francisco has significant implications for the AI industry and copyright law.

Judge William Alsup determined that Anthropic made “fair use” of books by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson to train its Claude large language model (LLM). The judge compared Anthropic's use of these books to a “reader aspiring to be a writer” who uses existing works to create something new and different.

However, the ruling also noted that Anthropic's copying and storage of more than 7 million pirated books in a central library did infringe on the authors’ copyrights. The company is now ordered to face a trial in December to determine the extent of damages owed for the infringement.

“Anthropic later bought a copy of a book it earlier stole off the internet will not absolve it of liability for the theft, but it may affect the extent of statutory damages,” Alsup wrote. US copyright law allows for damages of up to $150,000 per work for wilful copyright infringement.

The copyright issue has been a contentious point between AI firms and publishers. Generative AI models, such as the technology behind ChatGPT, require vast amounts of publicly available data to generate responses. Much of this data includes copyright-protected works.

An Anthropic spokesperson expressed satisfaction with the court's recognition that its AI training was transformative and aligned with copyright’s purpose of fostering creativity and scientific progress.

John Strand, a copyright lawyer at the US law firm Wolf Greenfield, emphasized the significance of the decision. “There are dozens of other cases involving similar questions of copyright infringement and fair use pending throughout the US, and Judge Alsup’s decision will be a crucial reference for those courts,” he said.

Given the number of ongoing AI copyright cases, it is expected that the primary question of whether training LLMs on copyrighted materials is fair use will eventually reach the US Supreme Court.

The writers filed a proposed class action against Anthropic last year, alleging the company used pirated versions of their books without permission or compensation to teach Claude to respond to human prompts. This case is one of many lawsuits brought by authors, news outlets, and other copyright owners against companies like OpenAI, Microsoft, and Meta Platforms over their AI training practices.

The doctrine of fair use allows the use of copyrighted works without the copyright owner’s permission in certain circumstances. This legal defense is crucial for tech companies, and Alsup’s decision is the first to address it in the context of generative AI.

AI companies argue that their systems make fair use of copyrighted material to create new, transformative content. They also contend that being forced to pay copyright holders could stifle the nascent industry. Anthropic told the court that its system copied the books to “study plaintiffs’ writing, extract uncopyrightable information from it, and use what it learned to create revolutionary technology.”

Giles Parsons, a partner at the UK law firm Browne Jacobson, noted that the ruling would have no impact in the UK, where the fair use argument is less applicable. Under current UK copyright law, copyright-protected work can be used without permission for scientific or academic research, but the UK’s fair use defense is much narrower.

Copyright owners in the US and UK argue that AI companies are unlawfully copying their work to generate competing content, threatening their livelihoods. A UK government proposal to change copyright law to allow use of copyright-protected work without permission, unless the owner opts out, has faced strong opposition from the creative industries.

Alsup also ruled that Anthropic violated the authors’ rights by saving pirated copies of their books as part of a “central library of all the books in the world.” Other prominent AI companies, including OpenAI and Meta, have also been accused of downloading pirated digital copies of millions of books to train their systems.

This decision highlights the complex interplay between AI innovation and copyright law, setting a precedent that will likely influence future legal battles in the tech industry.