Thomson Reuters Wins Appeal Case Against AI Company Over Fair Use

WISECODE Take

Thomson Reuters won its appeal against an AI company regarding fair use, a ruling that dampens the optimistic assumption among AI developers that training models automatically qualifies as fair use. The core of this lawsuit is whether an AI company can claim fair use when using copyrighted database contents without authorization for training. In this case, Thomson Reuters accused the AI company of copying its legal data, and the appeals court did not fully accept the AI company's fair use defense. This indicates that the judiciary is becoming increasingly strict regarding the sources of generative AI training data, no longer tolerating copyright infringement under the guise of technological innovation. For business owners developing their own AI models, this implies a significant increase in the licensing costs for training data in the future. Those who only implement established commercial AI tools without training their own foundational models remain unaffected. Businesses should take two actions: First, audit the training datasets used by R&D teams to ensure all sources have explicit commercial licenses. Second, when contracting with external AI vendors, explicitly require them to guarantee data compliance and include indemnification clauses to mitigate risks.