À PROPOS DE CET ÉPISODE
Speaker: Professor Bhamati Viswanathan, Visitor, Cambridge Law Faculty and Fellow at the Kernochan Center for Law, Media and the Arts at Columbia Law School
Biography: Bhamati Viswanathan is a Senior Visitor at the University of Cambridge Faculty of Law and a Fellow (Non-Resident) at the Kernochan Center for Law, Media and the Arts at Columbia Law School (New York). Prior to joining the Cambridge Faculty of Law, she was Assistant Professor at New England Law | Boston, where she taught copyright law, artificial intelligence and the law, law and the visual arts, intellectual property law, and U.S. Constitutional law. She is the author of “Cultivating Copyright: How Creative Industries Can Harness Intellectual Property to Survive the Digital Age” (Routledge/Taylor & Francis Press). She currently holds an Edison Fellowship from the Intellectual Property Policy Institute at University of Akron Law School, under whose aegis she is writing a series of articles on the disparate impact of copyright law on women creators and women-centric work. She is also planning a book on the nexus of intellectual property and arts/culture in the age of artificial intelligence.
Bhamati serves as Chair of the American Bar Association Intellectual Property Section: Visual and Dramatics Works Committee. She is a Faculty Advisor on the Copyright Alliance Academic Advisory Board. She serves as Faculty Partner to the News/Media Alliance. She is Education Advisor to the Volunteer Lawyers for the Arts (VLA)/ Massachusetts Arts and Business Council. She is also a Faculty Advisor to the Journal of the Copyright Society; and she was a Trustee of the Copyright Society, as well as Chair of its New England Chapter. She holds an S.J.D./LL.M. from University of Pennsylvania Law School; a J.D. from University of Michigan Law School; and a B.A. from Williams College. She is a competitive figure skater, violinist, and published poet/translator and lives in Boston.
Abstract: The training of generativeAI models on ingested work is a hotly contested area of U.S. copyright law. In this Seminar, I will inquire whether such training may constitute “fair use” under the nonexclusive four-factor test of the U.S. Copyright Act. Currently, courts are wrestling with the fair use defense in several major cases, including Thompson Reuters v. ROSS Intelligence; Bartz v. Anthropic; Kadrey v. Meta; and the consolidated litigation of In re: OpenAI.
Another open question is whether AI outputs infringe copyright in other works. Here, plaintiffs must establish that AI outputs infringe their works by passing the threshold of the “substantial similarity” test. I will discuss the test in the context of AI litigation, and will suggest that the relatively novel “market dilution” theory, focusing on harm caused by stylistically similar outputs, might be applied to weigh against a fair use defense for GenAI training. I will also address whether the theory of “vicarious liability” might be fruitfully brought to bear against certain genAI companies.
Lastly, I will ask what action Congress can, or should, take, with a view to striking a fair balance between meeting the needs of innovative technologies and securing the rights of creative industries and creators. As an example, I will raise a recent proposal (in which I was involved) that Congress explicitly prohibit GenAI training on materials derived from digital repositories of unlicensed materials (so-called “shadow libraries”).
For more information (and to download slides) see:
https://www.cipil.law.cam.ac.uk/seminars-and-events/cipil-seminars
