Machine Made, Legally Challenged: Large Language Models on Trial (Presented by The Federal Bar Association Annual Meeting & Convention 2025 sponsor Ciresi Conlin LLP)

This introductory session explores what large language models are and how they function, including their real-world applications in healthcare, finance, and consumer services. Attendees will learn how LLM development raises complex intellectual property questions around copyright, patents, and trademarks, setting the foundation for deeper legal analysis.

This comprehensive session examines the fair use doctrine and its four-factor balancing test as applied to LLM copyright cases. Through detailed analysis of landmark 2025 decisions including Thomson Reuters v. Ross Intelligence, Bartz v. Anthropic, and Kadrey v. Meta, attendees will understand how courts evaluate transformative use, market harm, and the critical distinction between training inputs and outputs. The session also covers the historic $1.5 billion Bartz settlement and emerging research on LLM memorization that may shape future litigation.

This session explores causes of action beyond traditional copyright claims and the significant barrier of copyright preemption that plaintiffs must overcome. Attendees will examine viable alternative theories including contract claims with express AI-training prohibitions, hot news misappropriation, trademark dilution, and emerging antitrust approaches like those in Helena World Chronicle v. Google. The session provides practical guidance on structuring terms of service and understanding which claims survive preemption analysis.