Anthropic Scores Major Victory in US Copyright Battle Over AI Training
According to US copyright law, statutory damages of up to $150,000 per work might be justified in cases of intentional copyright infringement.In a late Monday ruling, a federal judge in San Francisco upheld Anthropic’s use of unlicensed books to train an AI system, finding that the practice did not violate US copyright law.
A federal judge in San Francisco ruled late on Monday that Anthropic’s use of books without permission to train its artificial intelligence system was lawful under US copyright law. Siding with tech companies on a pivotal question for the AI industry, US District Judge William Alsup stated that Anthropic made “fair use” of books by writers Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson to train its Claude large language model. Alsup also stated that Anthropic’s copying and storage of over 7 million pirated books in a “central library” violated the authors’ copyrights and did not qualify as fair use. A trial has been ordered by the judge for December to ascertain the amount owed by Anthropic for the infringement.
Representatives for Anthropic have not yet provided a response to the request for comment regarding the ruling made on Tuesday.
The writers initiated a proposed class action against Anthropic last year, contending that the company, supported by Amazon and Alphabet, utilized unauthorized versions of their books without consent or compensation to train Claude to respond to human prompts. The proposed class action is among several lawsuits filed by authors, news outlets, and other copyright holders against companies such as OpenAI, Microsoft, and Meta Platforms regarding their AI training practices.
The doctrine of fair use permits the utilization of copyrighted works without obtaining permission from the copyright owner under certain conditions. Fair use serves as a crucial legal defense for tech companies, and Alsup’s ruling marks the first instance of addressing it within the realm of generative AI. AI companies contend that their systems utilize copyrighted material in a manner that is fair, generating new and transformative content. They assert that requiring payment to copyright holders for their work could stifle the growth of the emerging AI industry.
Anthropic informed the court that it utilized the books under fair use, asserting that US copyright law “not only allows, but encourages” its AI training as it fosters human creativity. The company stated that its system replicated the books to “study Plaintiffs’ writing, extract uncopyrightable information from it, and use what it learned to create revolutionary technology.” Copyright owners assert that AI companies are illegally reproducing their work to create rival content that jeopardizes their livelihoods.
Alsup concurred with Anthropic on Monday that its training was exceedingly transformative.”Like any reader aspiring to be a writer, Anthropic’s LLMs trained upon works not to race ahead and replicate or supplant them — but to turn a hard corner and create something different,” Alsup also stated, however, that Anthropic infringed upon the authors’ rights by storing pirated copies of their books as part of a “central library of all the books in the world” that would not necessarily be utilized for AI training.
Anthropic and other leading AI companies, such as OpenAI and Meta Platforms, have faced allegations of downloading pirated digital copies of millions of books for the purpose of training their systems. Anthropic informed Alsup in a court filing that the origin of its books was not pertinent to fair use. “This order questions whether any accused infringer can adequately justify why downloading source copies from pirate sites, which could have been purchased or accessed lawfully, was reasonably necessary for any subsequent fair use,” Alsup stated on Monday.
