A U.S. federal judge just handed artificial intelligence ("AI") one of its first big courtroom wins that will likely ripple through every major copyright case against AI companies.
In Bartz v. Anthropic, the court ruled that it is fair use to use legally purchased books to train an AI model. Judge Alsup called the use "quintessentially transformative," comparing teaching an AI how to write to a human reading a book to learn how to write. Since the AI wasn’t copying or republishing the books, just learning from them, he found it didn’t violate copyright laws.
This is huge for the growth of AI models. However, a real terrible moment for authors, artists, and other creators who have been arguing that using their work, especially without permission, amounts to large-scale theft.
This decision will shape the future, unless reversed on appeal. While Judge Alsup's ruling is not binding on other courts across the U.S., every founder, judge, and lawyer working on AI copyright issues needs to take notice.
To be clear, Judge Alsup's ruling doesn’t give AI companies a free pass. He has also allowed a separate claim to move forward about allegedly using pirated books for training. As a result, the how and where data comes from still matters.