Anthropic wins key US ruling on AI training in authors' copyright lawsuit
NORTHERN DISTRICT OF CALIFORNIA, UNITED STATES, JUN 24 – A federal judge ruled Anthropic's AI training on legally acquired books is transformative fair use but ordered a December trial over its use of more than 7 million pirated books.
- On Monday, US District Judge William Alsup determined that Anthropic’s training of its Claude AI model using millions of copyrighted books constitutes permissible fair use under US copyright law.
- The ruling followed a lawsuit by three writers accusing Anthropic of piracy for obtaining unauthorized digital copies from shadow libraries to build a central library of over seven million books.
- Judge Alsup emphasized that the training was 'exceedingly transformative' since Anthropic’s models do not replicate but create new content, and deemed the acquisition and scanning of physical books a reasonable practice.
- Alsup wrote it was unlikely any defendant could prove downloading pirated copies was necessary, adding Anthropic’s fair use defense aligns with copyright’s goal of fostering creativity as the authors did not allege replication or knockoffs.
- The ruling marks a significant win for AI development by validating transformative use but orders a trial to assess unlawful storage claims and means industry-wide licensing mandates might be avoided.
308 Articles
308 Articles
In a first-of-its-kind decision, an AI company wins a copyright infringement lawsuit brought by authors
The ruling opens a potential pathway for AI companies to train large language models on copyrighted works without authors' consent — but only if copies of the works were obtained legally.
Two judges in the US have given the right this week to Silicon Valley giants in AI cases. Anthropic won the most important: the use of millions of books to train their AI models comes within the allowed “legitimate use.” The judge of the case believes that Anthropic’s goal is “transformer” and is therefore not simply a copy of the original. It is the first big case that decides whether the training of AI models with protected material is legitim…
A U.S. judge ruled: AI can be dragged into a book protected by copyright. The decision could have effects in Canada.
Coverage Details
Bias Distribution
- 54% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium