AI and D - July 3, 2025
AI Training on Copyrighted Books - Fair or Unfair?
Two federal judges have delivered significant victories for AI companies in copyright lawsuits, ruling that training artificial intelligence models on copyrighted books constitutes fair use under U.S. law. These landmark rulings mark the first court decisions on AI training copyright disputes, setting crucial precedents for the tech industry.
Different Paths to Victory
Senior District Judge William Alsup ruled in Anthropic's favor on June 23, emphasizing the transformative nature of AI training. He called the technology "among the most transformative many of us will see in our lifetimes," finding that Claude's output didn't replace original works but created something new from them.
District Judge Vince Chhabria took a different approach in Meta's case on June 25, focusing on market harm rather than transformation. He ruled that the 13 plaintiffs, including Sarah Silverman and Ta-Nehisi Coates, failed to demonstrate that Meta's AI training substantially diminished the market for their original works.
Victory With Caveats
Both rulings come with important limitations. Judge Chhabria explicitly stated his decision doesn't mean "Meta's use of copyrighted materials to train its language models is lawful," describing the ruling as having "limited consequences" and essentially inviting stronger future challenges.
Neither company is completely clear yet. Both face separate allegations over illegally obtaining books from pirated databases, with Anthropic heading to trial over these piracy claims while Meta must negotiate with accusers.
Broader Industry Impact
These victories bolster fair use arguments across the tech industry, where dozens of similar cases against Google, OpenAI, Microsoft, and others remain pending. The outcomes could determine whether companies continue accessing free training data or must establish expensive licensing deals, potentially reshaping the entire AI landscape.
However, with over 40 pending cases and likely appeals, the legal battle is far from over. As lawyer Amir Ghavi notes, "there is still a long way to go before the issue is settled by the courts."
Key Points
First federal court rulings on AI training copyright disputes favor tech companies using different legal reasoning
Judge Alsup emphasized transformative nature of AI training; Judge Chhabria focused on lack of market harm evidence
Both companies still face separate trials over allegedly obtaining books from pirated databases
Rulings have limited scope and don't establish blanket protection for all AI training scenarios
Over 40 similar copyright lawsuits against major tech companies remain pending in courts
Cases could reshape AI industry economics by determining whether companies need expensive licensing deals