A U.S. District Court has ruled in favour of Thomson Reuters in its copyright infringement lawsuit against Ross Intelligence, a legal AI startup. The decision, issued on Tuesday by Judge Stephanos Bibas of the Delaware District Court, marks a significant moment in the ongoing debate over AI training and the use of copyrighted data. Filed in 2020, the case is among the first to address whether AI companies can use copyrighted content without permission under the “fair use” doctrine.
Key Takeaways from the Ruling
• Court rejects Ross Intelligence’s fair-use defence, ruling that its AI improperly used Thomson Reuters’ proprietary legal content. • Decision focuses on non-generative AI, meaning it does not yet set a direct precedent for large language models (LLMs) used by companies like OpenAI and Microsoft. • Ross shut down in 2021, citing financial struggles amid legal battles.
The copyright lawsuit was filed after Ross Intelligence used Thomson Reuters’ Westlaw database to train its AI-powered legal research search engine. While legal decisions themselves are not copyrightable, Westlaw’s headnotes—summaries of legal rulings written by human editors—are protected.
Judge Bibas ruled that Ross turned these headnotes into numerical data to train its AI, effectively replicating Thomson Reuters’ work without permission. The court deemed this copyright infringement and found no merit in Ross’s argument that the data was merely “added noise” in its AI training.
“None of Ross’s possible defences holds water,” Judge Bibas wrote in his ruling. He emphasized that Ross had created a direct competitor using Thomson Reuters’ proprietary content, which significantly impacted the market value of the original work.
This ruling could have major implications for AI companies, including OpenAI, Microsoft, and Google, which have all faced copyright lawsuits over the use of scraped data in training their AI models. While this case involved non-generative AI, it raises broader questions about whether LLMs trained on copyrighted material can claim fair use protection.
Currently, similar lawsuits are making their way through the courts, and many AI developers are closely watching how judges interpret AI’s relationship with copyrighted data.