Vai al contenuto

Questa traduzione dell'interfaccia è una bozza e non è stata rivista da una persona madrelingua. L'inglese è la versione di riferimento. English

Third Circuit upholds Thomson Reuters win over Ross Intelligence in first US appellate ruling on AI-training fair use

30 set 2026 · Raccolto automaticamente da fonti pubbliche. Non revisionato da una persona.

Questo contenuto è disponibile solo nella lingua originale (inglese).

What happened

The 3rd US Circuit Court of Appeals affirmed that defunct legal-research company Ross Intelligence did not make fair use of Thomson Reuters' copyrighted material to build an AI-based legal search engine. The opinion was unsealed on September 30, 2026.

Details

  • Thomson Reuters' 2020 lawsuit accused Ross of copying thousands of Westlaw headnotes (summaries of points of law in court opinions) to train a competing AI-based legal search engine. Ross shut down its platform in 2021, citing the costs of the litigation.
  • The appeals court said Ross's use of the headnotes to build a similar, competing product was not transformative, rejecting the fair use defense. It also rejected Ross's argument that the headnotes were not copyrightable, writing that the materials "possess a creative spark."
  • The court distinguished the case from other AI training disputes: Ross's engine did not feature generative AI, and the opinion notes in a footnote that concerns raised by the US Department of Justice in a copyright lawsuit against OpenAI "do not apply here."
  • The case is Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc, 3rd Circuit No. 25-2153, affirming a Delaware district court ruling. A unanimous three-judge panel wrote that "appearances can be deceiving" — despite Ross framing it as a case about the future of AI legal technology, "in truth, this is no more than an ordinary copyright case."

Boundary

This is a ruling about a non-generative, competing legal-search tool. The court expressly carved out generative AI, so the holding should not be read as settling fair use for generative model training.

Fonti