Anthropic wins US copyright lawsuit

But will still face trial for using “shadow libraries”

A US judge has ruled that Anthropic did not break the law by training Claude on millions of copyrighted books.

In a test case for Big AI, a US judge has dismissed the claim of copyright infringement from several authors by ruling that Anthropic did not break the law by training its LLM Claude on millions of copyrighted books.

However, Anthropic must stand trial later this year for the way it got hold of those books which involved downloading them from online “shadow libraries” of pirated copies.

US District Judge William Alsup of San Francisco said in a ruling that Anthropic’s system's ingestion of the work of hundreds of thousands of writers was “fair use” under US copyright law because it was “quintessentially transformative.”

“Like any reader aspiring to be a writer, Anthropic’s (AI large language models) 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 matter isn’t fully settled because Alsup also said Anthropic must still go to trial in December because “Anthropic had no entitlement to use pirated copies for its central library,”

Documents disclosed in court showed that Anthropic started to express doubts about the legal implication of using shadow libraries as it came under scrutiny and subsequently started buying copies of digital books. This cut little ice with Judge Alsop, who wrote:

“That 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.”

The ruling is the latest twist in the ongoing battle between creatives and the makers of GenAI. The UK government is currently considering options for a copyright opt-out for creatives – a change from its previous position which was that having to seek permission from authors would stifle innovation.

OpenAI and Meta are both facing lawsuits from copyright owners in the US, but these are winding their way through the legal system at a much slower pace than the LLMs are developing. Big AI has taken the traditional Big Tech mindset that it’s better to ask for forgiveness than permission.

Judge Alsop’s ruling suggests that this forgiveness may not be forthcoming – at least not entirely.

Anthropic said Tuesday it was pleased that Judge Alsop recognised the transformative nature of Claude’s training and that this was consistent with “copyright’s purpose in enabling creativity and fostering scientific progress.”