US Appeals Court Upholds Thomson Reuters' Landmark AI Copyright Win

A US appeals court just delivered a landmark ruling, upholding Thomson Reuters' victory in its high-stakes copyright battle against former rival Ross Intelligence. This decision critically reinforces the idea that using copyrighted content to train AI models can indeed be considered infringement, sending a clear message across the rapidly evolving artificial intelligence landscape. The dispute, which began in 2020, centered on Ross Intelligence alleged 'scraping' of Thomson Reuters' extensive Westlaw legal database to build its own AI-powered legal search engine. Ross Intelligence ceased operations in January 2021, partly due to the financial strain of this very lawsuit. This latest ruling by the 3rd US Circuit Court of Appeals is a massive win for traditional content owners, validating their efforts to protect intellectual property against unchecked AI training practices. The court explicitly rejected Ross's 'fair use' defense, a pivotal aspect of the case. While this specific case involved a non-generative AI tool, the implications are significant for other ongoing copyright lawsuits against developers of generative AI. This appeals court decision is a game-changer, potentially shaping how AI developers acquire and license data going forward. It suggests that clear permissions and licensing agreements will become non-negotiable for training large language models (LLMs), moving away from a 'move fast and break things' approach. Expect more vigorous enforcement of intellectual property rights and possibly new industry standards emerging for data sourcing. Companies relying on freely available web data for AI training should be re-evaluating their strategies right now, as the legal risks just significantly ratcheted up.