Publicly Detectable Watermarking for Language Models - PPML 2024 Invited Talk

Explore a cutting-edge invited talk from the Privacy-Preserving Machine Learning Workshop (PPML) 2024, an affiliated event of Crypto 2024. Delve into Sanjam Garg's presentation titled "Publicly Detectable Watermarking for Language Models," chaired by Harish Karthikeyan. Gain insights into the latest advancements in privacy-preserving techniques for machine learning, focusing on watermarking methods for language models. Learn about the importance of public detectability in watermarking and its implications for protecting intellectual property in AI models. Discover how this research contributes to the broader field of cryptography and machine learning security during this 50-minute talk, which serves as the first of three invited presentations at the workshop.

PPML 2024 invited talk I by Sanjam Garg