US Judge Rules in Favour of AI Firm in Copyright Dispute Over Book Training Data
A U.S. federal judge has ruled that artificial intelligence company Anthropic did not violate copyright law by using published books to train its AI models, marking a significant development in the growing legal debate around AI training practices.
The lawsuit, filed last year by three authors – including mystery novelist Andrea Bartz – accused Anthropic of unlawfully copying their copyrighted books to develop its Claude AI model. Bartz, along with non-fiction writers Charles Graeber and Kirk Wallace Johnson, claimed the tech firm exploited their work to build a multi-billion-dollar business.
In his ruling, Judge William Alsup determined that Anthropic’s use of the books was “exceedingly transformative,” meaning the material was used in a way that significantly altered its original purpose – a key standard under U.S. copyright law’s fair use doctrine.
“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,” the judge wrote.
However, the court rejected Anthropic’s bid to dismiss the case outright. While the training process itself was deemed fair use, Judge Alsup ruled that a trial must still be held to examine how the company acquired the training data. Specifically, Anthropic is alleged to have stored over seven million pirated books in what the court described as a “central library of all the books in the world.”
The ruling is among the first of its kind to address how AI firms can use copyrighted works to train large language models (LLMs), a matter at the heart of many ongoing lawsuits across the creative industries.
Anthropic, which is backed by Amazon and Google parent company Alphabet, faces potential damages of up to $150,000 per copyrighted work if found liable for improper data acquisition or storage.
In a statement, the company said it was pleased that the court recognised its use of the books as transformative but expressed disappointment over the decision to proceed to trial on the data acquisition claims.
“We remain confident in our case and are evaluating our options,” Anthropic said.
The authors’ legal team declined to comment.
This case is part of a broader wave of legal challenges facing AI firms over their use of copyrighted content. Earlier this month, Disney and Universal filed suit against AI image generator Midjourney, while major news organisations, including the BBC, are reportedly considering legal action over unauthorised use of their material in AI development.
Some AI companies have responded to the legal uncertainty by securing licensing deals with content creators and publishers to avoid future disputes.
Judge Alsup’s decision – allowing fair use but insisting on accountability for how the data was obtained – could influence how courts approach similar cases in the months ahead.
