US Third Circuit Landmark Ruling: Training AI on Copyrighted Data is Not 'Fair Use'

On September 30, 2026, the US Court of Appeals for the Third Circuit issued a landmark decision: a three-judge panel unanimously affirmed that AI startup ROSS Intelligence infringed copyrights by scraping content from Thomson Reuters' Westlaw legal database to train its AI search platform, explicitly rejecting the startup's fair use defense. This marks the first time a US federal appeals court has directly ruled on whether commercial AI model training qualifies as fair use under copyright law.

Writing for the panel, Circuit Judge Tamika Montgomery-Reeves stated: "Under ROSS' framing, this case appears to concern the future of AI legal technology. But appearances can be deceiving. In truth, this is no more than an ordinary copyright case." The court found that Westlaw's editorial headnotes possess the requisite "creative spark" to warrant full copyright protection, and that ROSS's use of them for AI training was "minimally transformative at best," directly harming both Westlaw's primary market and its potential derivative market for training data licensing.

For years, leading AI companies including OpenAI, Anthropic, and Meta have relied on the argument that ingesting copyrighted text and media is transformative fair use as their core legal shield against hundreds of lawsuits by authors, visual artists, and publishers. The Third Circuit's binding precedent delivers a major blow to that defense across active federal litigation. Defense counsel Yar Chaikovsky of White & Case LLP indicated ROSS intends to petition the US Supreme Court for review.

Sources: Courthouse News Service, Reuters Legal, Law.com.

BenPig clenches its tiny fists tightly, thinking: so taking what other people wrote to feed a machine really doesn't count as your own work.