Thomson Reuters has secured a significant legal victory in its copyright lawsuit against Ross Intelligence, a legal tech company. A U.S. federal judge ruled that Ross Intelligence infringed on Thomson Reuters' intellectual property by using content from its Westlaw legal research service to train its AI platform. This decision could influence over 39 other AI-related copyright lawsuits currently in U.S. courts, although it doesn't guarantee success for all plaintiffs claiming IP infringement by AI companies.
Ross Intelligence was accused of using Westlaw's case summaries to train its AI, which is marketed as a tool for analyzing documents and performing query-based searches in court files. Ross argued that its use of the copyrighted summaries was legally defensible as transformative, meaning it repurposed the summaries for a distinctly different function or market. However, Judge Stephanos Bibas found this argument unconvincing, stating that Ross's platform merely replicated Westlaw's service without adding new meaning or purpose.
The ruling also noted Ross's commercial intent, which undermined its defense. The company aimed to profit from a product directly competing with Westlaw without significantly recontextualizing the copyrighted material. According to Syracuse University law professor Shubha Ghosh, this represents a major win for Thomson Reuters, although the case will continue to trial.
While some plaintiffs in other AI copyright cases have sought to reference this decision, its broader applicability remains uncertain. The ruling specifically distinguishes between generative AI, which creates new content, and the type of AI used by Ross. Generative AI companies, like OpenAI, argue that fair use protects their data scraping for training, but not all copyright holders agree.
