Experience Stern | Faculty & Research

Overview: In a new paper, “Rethinking Intellectual Property Law in an Era of Generative AI,” NYU Stern Professor Arun Sundararajan, along with co-author Giuseppina D’Auria, explore how the rapid advances in generative Artificial Intelligence (AI) will shape the future of copyright and intellectual property (IP) protection, highlighting critical issues that policymakers must consider when enacting legislation designed to regulate artificial intelligence.

Why study this now: The recent copyright infringement lawsuit filed by The New York Times against OpenAI and Microsoft is the latest in a series of legal challenges to generative AI. Akin to during the rise of the Internet, society faces a pivotal moment, as AI distorts how the ownership of current and future content and innovation is allocated to creators, consumers, and future innovators. A key question that policymakers must grapple with: should new policy grant greater rights to human creators, potentially slowing digital innovation, or grant no IP rights to a human over what they consider their highly individual artistic style, which could have a chilling effect on both human creativity and broader societal innovation?

Top takeaways: 

• The authors identified three questions central to the current debate:
       1. What control does the owner of data have over its use in training a generative AI system?
       2. Does a human have the right to control the creation of a generative AI system that replicates their individual "creative process," and if such a system is created, what claims or recourse does the human have? 
       3. Who owns the works created by such a system?
• Drawing on examples of recently released AI-generated music and art, the authors make a case that a substantial gap exists between IP protections afforded by existing case law and what may be in society's best interest, and thus, the emergence of generative AI technologies necessitate the most substantive change to IP law since the Copyright Act was passed.
• The research notes that there is a little consensus among experts about the future trajectory of generative AI capability. Moreover, it is unclear whether AI systems trained on AI-generated content will continue to "learn" new capabilities at the same rate (or at all), adding another layer of uncertainty and making it possible that the systems need human-generated training data in the mix.

Key insight: “If left unchanged, it is more likely than not that the current IP regime will favor a dramatic shift away from human-led creation and towards one where more and more works are generated by machines,” said the authors. “Thus, a key question society must consider is whether it is comfortable with a vast majority of future creation being done by machines rather than humans, and policy makers must in addition consider whether a future of primarily AI-generated creative works is on the optimal path.”

The research was published in the TechREG Chronicle in November 2023.