Key points:
- The Third Circuit is weighing whether ROSS Intelligence infringed Thomson Reuters’ copyright by training AI on Westlaw headnotes.
- At least 20 amicus briefs highlight sector-wide stakes for legal data access and AI innovation.
- The outcome could shape U.S. copyright doctrine for AI training and influence future legislative reforms.
A closely watched dispute now before the Third Circuit Court of Appeals could redefine how copyright law governs the training of artificial intelligence systems. The case, detailed in a recent Law.com report, centers on whether ROSS Intelligence unlawfully used Thomson Reuters’ Westlaw headnotes to train its legal research algorithms.
At least 20 amicus briefs have been filed, underscoring the breadth of interest across the legal, technology and media sectors. The dispute presents a critical test: does training an AI model on editorialized legal content—such as headnotes summarizing judicial opinions—constitute copyright infringement?
Supporters of ROSS, including several legal tech companies, argue that restricting training data access would impede innovation in legal research. They contend that headnotes function as descriptive metadata and that preventing their use risks entrenching incumbents and stifling competitive development of more efficient research tools. The argument echoes long-running debates about interoperability and access in legal publishing.
Thomson Reuters and aligned media organizations frame the issue differently, asserting that uncompensated reuse of curated editorial materials undermines the economic incentive to invest in high-quality legal content. Their position extends beyond law libraries: sectors reliant on proprietary data sets worry that a ruling favorable to ROSS could weaken intellectual-property protections for materials used to train a growing class of AI systems.
The case’s implications are broad enough that legal experts anticipate its influence could extend to future legislation. As noted in analysis referenced by Law.com, lawmakers may face pressure to clarify how copyright doctrines such as fair use apply to machine learning workflows. With AI systems increasingly embedded in legal practice, doctrinal uncertainty carries operational and compliance risks for corporate legal departments and law firms alike.
For now, the Third Circuit’s decision will serve as a bellwether. A ruling for Thomson Reuters could reinforce strong protections for curated legal content, potentially raising costs and limiting training-data access for emerging AI tools. A ruling for ROSS, by contrast, may accelerate development of open-data models but unsettle longstanding assumptions about the proprietary value of editorial work product.
