Large Libel Models? When Do AI Hallucinations Become Defamation?
When an AI system fabricates damaging claims about a real person, is the AI company liable for defamation? Just last week, in Keene v. Google, a federal court confronted this question for the first time, holding that allegedly false Google AI search summaries could be viewed as potentially defamatory factual assertions – and that “actual malice” on Google’s part could be shown if Google’s AI kept outputting the falsehoods after plaintiff had expressly alerted Google about this. In July, a Delaware state trial court likewise allowed conservative activist Robby Starbuck’s defamation suit against Google to proceed.
What do these early decisions mean for defamation law in the age of large language models? Join us as we discuss when AI-generated falsehoods may become actionable, how traditional libel doctrines apply to AI outputs, and what these cases could mean for AI developers, users, and the future of online information.
Featuring:
- Prof. Lyrissa Lidsky, Raymond & Miriam Ehrlich Eminent Scholar Chair in US Constitutional Law, University of Florida Levin College of Law
- Prof. Eugene Volokh, Thomas M. Siebel Senior Fellow, The Hoover Institution, Stanford University; Gary T. Schwartz Distinguished Professor of Law Emeritus, UCLA School of Law
- (Moderator) Ben Daus, Strategic Counsel, Cognition AI
*******
As always, the Federalist Society takes no position on particular legal or public policy issues; all expressions of opinion are those of the speaker.
