Anthropic Wins Partial Victory in AI Copyright Dispute

AI firm Anthropic has secured a significant legal victory in a copyright dispute over the use of copyrighted material to train its AI models. U.S. District Judge William Alsup ruled that Anthropic’s use of copyrighted books to train its AI chatbot, Claude, qualifies as “fair use” under U.S. copyright law. The judge noted that the AI training process is “exceedingly transformative,” as Claude’s outputs do not replicate or regurgitate authors’ works but generate new text that is distinct from the originals.

However, the court also found that Anthropic’s practice of storing millions of pirated books in a permanent library constitutes a clear violation of copyright law. The judge emphasized that there is no exception in the Copyright Act for AI companies, and maintaining a permanent library of stolen works could destroy the academic publishing market if allowed.

The case was brought by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, who accused Anthropic of building Claude using millions of pirated books downloaded from sites like Library Genesis and Pirate Library Mirror. The lawsuit seeks damages and a permanent injunction, alleging that Anthropic built a multibillion-dollar business by stealing hundreds of thousands of copyrighted books.

Court records revealed that Anthropic downloaded at least seven million pirated books, including copies of each author’s works, to assemble its library. Internal emails showed that Anthropic co-founders sought to avoid the “legal/practice/business slog” of licensing books, while employees described the goal as creating a digital collection of “all the books in the world” to be kept “forever.”

Judge William Alsup’s ruling is the first substantive decision by a U.S. federal court that directly analyzes and applies the doctrine of fair use specifically to the use of copyrighted material for training generative AI models. The court distinguished between copies used directly for AI training, which were deemed fair use, and the retained pirated copies, which will now be subject to further legal proceedings, including potential damages.

While several lawsuits have been filed against other AI companies, including high-profile cases against OpenAI and Meta, those cases are still in early stages, with motions to dismiss pending or discovery ongoing. OpenAI and Meta both face lawsuits from groups of authors alleging their copyrighted works were exploited without consent to train large language models such as ChatGPT and LLaMA. The New York Times also sued OpenAI and Microsoft in 2023, accusing them of using millions of Times articles without permission to develop AI tools.

Reddit recently sued Anthropic, alleging it scraped Reddit’s platform over 100,000 times to train Claude, despite claiming to have stopped. This ongoing legal battle highlights the complex issues surrounding the use of copyrighted material in AI training and the need for clear guidelines and regulations in this rapidly evolving field.