A federal judge in Delaware, Stephanos Bibas, has decided that the copyright lawsuit filed by Thomson Reuters against Ross Intelligence must be resolved by a jury. Thomson Reuters has alleged that Ross unlawfully copied material from its legal research platform, Westlaw, to train an AI-based platform that competes with it. This ruling paves the way for one of the initial trials focused on the unauthorised utilisation of data for AI training.

Thomson Reuters claims that Ross Intelligence improperly employed its proprietary commentary, analysis, and organisational system, particularly the 'headnotes' summarising legal concepts in court judgments. Ross countered by asserting that it fairly used Westlaw material to locate judicial decisions, rather than directly competing with the content.

Judge Bibas concluded that a jury should resolve issues like fair use and the extent of Thomson Reuters' copyright protection for the headnotes. He also underscored the ongoing debate regarding whether training AI with copyrighted material serves the public interest.

Why does it matter?

The debate over whether AI should be trained with copyrighted material raises broader questions about the public interest in innovation versus intellectual property protection. Numerous technology companies, including Meta Platforms, Stability AI, and Microsoft-backed OpenAI, are currently confronting similar legal actions concerning the utilisation of copyrighted content for training their AI systems. The outcome should impact the way companies develop AI technologies, determining the boundaries of fair use and copyright protection in AI training data.