Open Technical Questions in Generative AI Copyright and Risk Mitigation

Watch a 51-minute lecture from the Simons Institute where Google researcher Yangsibo Huang explores critical technical challenges surrounding copyright issues in generative AI. Delve into the quantification and analysis of factors that may trigger copyright risks in GenAI models, and examine potential technical solutions for mitigating these risks. Learn about the effectiveness of various risk mitigation measures available to model deployers, and discover potential approaches for fairly compensating copyright owners in cases of confirmed infringement. Gain insights into the technical dimensions of the ongoing debate between generative AI companies and copyright holders, moving beyond legal fair use discussions to focus on practical implementation challenges and solutions.