AI Trained on Famed Authors’ Copyrighted Work. They Want Revenge

Generative AI tools for public use are a relatively new development, but they’re already promising to change the world. Systems like OpenAI’s ChatGPT can answer complex questions, write poems and code, and even mimic famous authors with uncanny accuracy. But this raises the question of whether using copyrighted materials to train these powerful AI products infringes the rights of the creators of the work?

This season on UnCommon Law, we’ll explore the intersection between artificial intelligence and the law. In the first two episodes, we examine how large language models actually ingest and learn from the billions of online data points, including copyrighted works. That training allows generative AI systems to respond to prompts and generate human language so adeptly. And we explore the lawsuits filed by creators who claim their copyrights were exploited without permission to feed the data-hungry algorithms powering tools like ChatGPT.

Can AI companies claim a fair use exception for their all-consuming training? Or could they be compelled to obtain licenses and pay to use the books, articles, screenplays, poems, codes, and prose that make the machine such a powerful tool?

Guests:

• Matthew Butterick, founder at Butterick Law, and co-counsel with the Joseph Saveri Law Firm on class-action lawsuits against OpenAI and others
• Isaiah Poritz, technology reporter for Bloomberg Law
• James Grimmelmann, professor of digital and information law at Cornell Tech and Cornell Law School
• Matthew Sag, professor of law and artificial intelligence, machine learning and data science at Emory University School of Law
• Mark Lemley, professor of law at Stanford Law School and the director of the Stanford Program in Law, Science and Technology, who is also representing Meta and Stability AI in the copyright cases against them