Judge Declares Copyrighted Books as Fair Use for AI Training Purposes

A federal judge has ruled in favor of Anthropic in a significant copyright case, stating that artificial intelligence developers can train their models on published books without needing consent from the authors. This ruling, issued by Judge William Alsup in the U.S. District Court for the Northern District of California, establishes a precedent that training AI on copyrighted material falls under the category of fair use. The decision addresses a long-standing concern among creatives across various fields, particularly as generative AI tools have gained popularity, enabling users to create art and content based on models that may have utilized copyrighted material without the consent of original creators.

Since 2023, numerous copyright lawsuits have been filed against AI companies by media firms, music labels, and authors. Artists have also expressed their concerns through open letters, calling for stricter regulations on the unauthorized use of their works. In response, some companies have begun negotiating licensing agreements with AI developers to clarify how their content can be used.

The lawsuit brought forth by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson claimed that Anthropic breached copyright laws by digitally copying millions of books to train its large language models, which generate humanlike text. Judge Alsup stated that the use of these copies was justified under fair use, emphasizing that most factors except for the nature of the copyrighted works favored this conclusion. He argued that the technology involved was among the most transformative of our time.

Despite the ruling supporting Anthropic’s position on fair use, Judge Alsup noted that the company still needs to address the allegations of using pirated copies in its extensive library for training its AI models. He clarified that purchasing books after illegally downloading them does not absolve Anthropic of liability for copyright infringement, although it might affect potential damages in the case.