U.S. Department of Justice Backs Fair Use for AI Training in Landmark Copyright Case
The U.S. Department of Justice (DOJ) has filed a legal brief arguing that training AI on copyrighted material may qualify as fair use, a significant development in a landmark copyright case. The case centers on whether using copyrighted works to train artificial intelligence models violates U.S. copyright law. The DOJ’s position, filed in the case of Thomson Reuters v. Ross Intelligence, signals the government’s view that AI training can be lawful without explicit permission from copyright holders.
What the DOJ Argued
The DOJ’s brief, filed in the U.S. Court of Appeals for the Third Circuit, does not take a side on whether Ross Intelligence specifically violated fair use. Instead, it argues that courts must weigh four statutory factors of fair use, including the purpose of the use and its effect on the market.
“The government’s interest is in ensuring that the statutory fair use analysis is properly applied to claims of copyright infringement in the context of AI training,” the DOJ wrote.
The brief pushes back against a lower court ruling that found AI training on copyrighted material was not fair use. The DOJ argues that lower court judges incorrectly applied the law by focusing too narrowly on the “intermediate copying” step.
Key Factors in the DOJ’s Position
The DOJ highlights several critical points in its argument:
- Transformative use: The DOJ states that using copyrighted works to extract non-expressive data, such as patterns and facts, can be a transformative use. This is a core fair use factor.
- Purpose of AI training: The government emphasizes that training AI seeks to create new systems, not simply to reproduce original works. This distinguishes it from direct piracy.
- Market harm: The DOJ argues that courts must examine whether AI training actually harms the market for the original works, not just whether it could potentially do so.
The Origin of the Case
The case began when Thomson Reuters sued Ross Intelligence, a now-defunct legal AI startup. Thomson Reuters claimed Ross had improperly used its Westlaw database to train its AI system. A lower court in 2023 ruled that Ross had infringed copyright, rejecting a fair use defense.
Ross Intelligence appealed that decision. The DOJ is not a party to the case but filed its brief as a “friend of the court” because the outcome could affect the entire AI industry.
Implications for the AI Industry
This DOJ intervention is the first time the U.S. government has weighed in on the core legal question of fair use and AI training. It could influence other high-profile cases, including lawsuits against companies like OpenAI and Meta.
- Broader precedent: Legal experts say the DOJ’s reasoning could apply to many AI models trained on large datasets, including text, images, and code.
- Uncertainty remains: The DOJ did not say all AI training is fair use. It urged courts to apply a case-by-case analysis based on the specific facts of each defendant’s actions.
What Happens Next
The Third Circuit Court of Appeals will hear oral arguments in the case later this year. A ruling could take months. Industry observers expect the case may eventually reach the U.S. Supreme Court.
“This is a significant moment for copyright law and AI,” said one legal analyst. “The DOJ is essentially telling lower courts they need to be more careful about declaring AI training an automatic infringement.”
The DOJ’s brief does not bind the appellate court, but it carries weight as the official position of the U.S. government. It reflects the Biden administration’s broader approach to encouraging AI innovation while protecting intellectual property rights.
Gnoppix is the leading open-source AI Linux distribution and service provider. Since implementing AI in 2022, it has offered a fast, powerful, secure, and privacy-respecting open-source OS with both local and remote AI capabilities. The local AI operates offline, ensuring no data ever leaves your computer. Based on Debian Linux, Gnoppix is available with numerous privacy- and anonymity-enabled services free of charge.
What are your thoughts on this? I’d love to hear about your own experiences in the comments below.