In this talk, I intend to explore various ways in which intellectual property law can be adapted in order to accommodate the challenges posed by AI, and what its underlying structure reveals about the legal system’s conception of agency, control, and responsibility. After briefly reviewing the current state of the debate concerning copyright and patent eligibility for AI-assisted outputs, I propose a test with which to apply the principles of the existing doctrine to AI-related cases. I argue that the proposed test is consistent with the broader logic of legal responsibility across various domains of law, be it with respect to obligations (e.g., tort liability) or entitlements (e.g., IP rights), and that it tracks an agent's ability to envision and shape specific outcomes.
From there, I reflect on what IP's (and the legal system's) implicit theory of agency suggests about the kinds of responsibility AI systems can and cannot bear, and sketch a broader proposal for grounding both fault-based and strict liability in the relationship between an agent's choice of outcomes and the agent's control over the risks associated with those outcomes. The proposed framework could provide a more principled basis for strict liability and help explain why strict liability for harms arising from AI systems with high levels of autonomy and/or emergent behavior is justified.
I conclude by briefly considering how IP's treatment of identity and expression could help address other pressing challenges in AI policy—for example, using the concept of appropriation of likeness to prohibit the nonconsensual use of personal data in training generative models that produce deepfake pornography, even when the individuals concerned are not directly or recognizably depicted in the output.
Jonathan Iwry is a Fellow at the Accountable AI Lab at the Wharton School of the University of Pennsylvania. His work focuses on the philosophical challenges posed by AI and other emerging technologies for foundational legal concepts. He has authored or co-authored scholarly work on topics that include machine learning in psychological language analysis, the ethics of noninvasive brain stimulation, the legal implications of the metaverse, FDA’s use of its emergency powers during the COVID-19 pandemic, behavioral economics in drug policy, and the law of outer space. His work has been published in the Food and Drug Law Journal, Bloomberg Law, Psychological Methods, Frontiers in Human Neuroscience, the Negotiation Journal (of Harvard's Program on Negotiation), and the Oxford Handbook of Secularism. He was previously a corporate associate at the law firm Ropes & Gray LLP. He received his J.D. from Harvard Law School and B.A., summa cum laude, in Philosophy and History from the University of Pennsylvania. During law school, he served as a Teaching Fellow in two Harvard College courses—Michael Sandel’s “Justice” course and Joshua Greene’s course on AI ethics—and received awards from the university for excellence in teaching. He is also a Technology, Law, and Policy Fellow at the Center for the Future of AI, Mind & Society at Florida Atlantic University. He moonlights as a freestyle rap artist and is an eleven-time winner of the Supreme Bars rap tournament in Brooklyn.
