Radio News Now — AI Legal Watch
A federal appeals court has handed copyright owners a major victory in a case involving the use of protected material to train an artificial intelligence system — but the ruling does not establish that all AI training on copyrighted works is unlawful.
On September 29, 2026, the U.S. Court of Appeals for the Third Circuit affirmed a lower-court ruling in favor of Thomson Reuters and West Publishing in their copyright case against Ross Intelligence. The dispute centered on Ross’ use of copyrighted Westlaw headnotes while developing an AI-powered legal research platform.
The Third Circuit held that the Westlaw headnotes at issue were sufficiently original to receive copyright protection and concluded that Ross’ use of them was not protected by the fair-use doctrine.
Why the Decision Matters
The case is significant because it is the first federal appellate decision addressing fair use in the context of AI training. But the court went out of its way to distinguish Ross’ technology from generative AI systems such as large language models.
Ross’ system did not generate original expression. Instead, according to the court, Ross used Thomson Reuters’ headnotes to help create a competing legal-research service that served a highly similar commercial purpose.
That distinction matters. The opinion should not be read as a blanket ruling that training a generative AI model on copyrighted material automatically violates copyright law.
The Licensing-Market Question
One of the most consequential parts of the decision involves the fourth fair-use factor: the effect of the use on the potential market for the copyrighted work.
The court rejected the argument that there could be no market harm simply because Thomson Reuters had not yet licensed its headnotes specifically as AI-training data. The judges concluded that unauthorized copying could interfere with a copyright owner’s opportunity to enter a derivative licensing market in the future.
That reasoning could become important well beyond legal research.
Publishers, broadcasters, music companies, photographers, writers and other rights holders are increasingly asking whether AI developers should obtain licenses before using copyrighted material for model development. The Third Circuit’s ruling gives copyright owners a stronger argument that a potential AI-training licensing market can have economic value even before that market is fully established.
What the Court Did Not Decide
The ruling does not resolve the much larger legal battle over generative AI.
The Third Circuit specifically noted that Ross’ platform differed from generative systems capable of producing original responses. The court also distinguished pending litigation involving large language models, where judges may have to weigh different facts concerning transformation, market substitution and the nature of the resulting output.
That means the decision is important precedent — but not the final word on whether companies such as OpenAI, Anthropic, Google, Meta or other generative-AI developers may lawfully train models on copyrighted material without permission.
RNN Bottom Line
The bigger question is no longer simply whether copyrighted material was copied.
The WHY is whether an AI company can use copyrighted material without permission in a way that competes with the copyright owner or bypasses a legitimate licensing market.
In the Ross case, the Third Circuit said the answer was no.
For radio, music and digital media companies, the decision is another reason to pay close attention to licensing rights, content archives, syndicated material, voice recordings, music catalogs and other intellectual property that could have value as AI-training data.
Ross Intelligence has said it intends to seek review by the U.S. Supreme Court.
Sources: U.S. Court of Appeals for the Third Circuit, Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc., No. 25-2153; Reuters reporting dated September 29–30, 2026.