A federal court ruled that the training of an AI tool on copyrighted materials was ineligible for a “Fair Use” defense for its copying. In this video, Bryan Sterba informs viewers on key takeaways from the judge’s decision, and what it could mean for developers and users of generative AI tools.
Speakers:
Bryan Sterba, Partner, IP & Tech Transactions, Lowenstein AI, and Emerging Companies & Venture Capital
Lowenstein AI: A-I Didn’t Know That
Partner Bryan Sterba examines the myriad legal and business implications of this revolutionary technology. Bryan and guests from across firm practices discuss the many ways that businesses can leverage AI to increase productivity and efficiency, as well as the challenges presented as companies, legislatures, and regulatory agencies weigh differing approaches to managing intellectual property rights, privacy and data security, and national security interests, amongst other concerns.
READ THE TRANSCRIPT
Bryan Sterba: Hi, I'm Bryan Sterba, a partner in Lowenstein Sandler's AI practice here with another episode of "AI Didn't Know That."
The U.S. District Court for the District of Delaware recently held in the case of Thomson Reuters vs. Ross Intelligence that defendants could not rely on a "fair use" defense when it came to their copying of Thomson Reuters’ content to train Ross's AI tool.
Now, Ross is a startup that created an AI tool and platform to optimize case law research based on the user's queries. Think of it as an AI powered search engine for lawyers looking to get answers about current case law.
Thomson Reuters, the owner of Westlaw, filed a lawsuit against Ross, alleging that it infringed its by using a data set copied from Westlaw's head notes and case summaries to train its AI tool.
This ruling is the first reported case to address the critical question of fair use in AI training. In this case, the factor that most heavily influenced the decision against applying fair use was the purpose and character of the usage. Here, the fact that Ross's intent was to develop a competing tool made it extremely difficult to apply the fair use defense.
Content publishers will surely point to this decision in asserting claims against generative AI developers and users when those AI tools are used to generate content that displaces the market for their own. But this decision explicitly disclaimed its applicability in the context of generative AI, with the judge noting for readers that only non-generative
AI is before the court today.
So, whether using copyrighted materials to train generative AI tools qualifies as fair use remains an open question.
We'll need to wait for other decisions in the many cases we've been following relating training of generative AI on such content. The decision here is instructive in that it follows the same theme that other copyright cases have focused on, which is that the most important factor in determining whether a certain act of copying is fair use will be the impact on the market for the copied works.
Either way, we can certainly expect plaintiffs to cite this decision when usage affects the market for their copyrighted materials and defendants to point to its applicability in the context of generative AI.
For more information, please read our client alert linked below, and stay tuned for our analysis on these other cases as decisions develop.
Thank you for watching “AI Didn't Know That.”
