no code implementations • 27 Nov 2024 • Xuandong Zhao, Sam Gunn, Miranda Christ, Jaiden Fairoze, Andres Fabrega, Nicholas Carlini, Sanjam Garg, Sanghyun Hong, Milad Nasr, Florian Tramer, Somesh Jha, Lei LI, Yu-Xiang Wang, Dawn Song
As the outputs of generative AI (GenAI) techniques improve in quality, it becomes increasingly challenging to distinguish them from human-created content.
no code implementations • 24 Oct 2024 • Miranda Christ, Sam Gunn, Tal Malkin, Mariana Raykova
We demonstrate robustness to both token substitution and perturbation of the model parameters.
1 code implementation • 9 Oct 2024 • Sam Gunn, Xuandong Zhao, Dawn Song
Our experiments verify that, in contrast to every prior scheme we tested, our watermark does not degrade image quality.
no code implementations • 14 Feb 2024 • Miranda Christ, Sam Gunn
As our primary application of pseudorandom codes, we present an undetectable watermarking scheme for outputs of language models that is robust to cropping and a constant rate of random substitutions and deletions.
no code implementations • 25 May 2023 • Miranda Christ, Sam Gunn, Or Zamir
We ask: Is it possible to introduce a watermark without incurring any detectable change to the output distribution?