Watermarking Autoregressive Image Generation

TL;DR

This paper introduces a novel method for watermarking autoregressive image generation outputs at the token level, adapting language model techniques to ensure robust and reliable detection against various attacks.

Contribution

It presents the first approach to watermark autoregressive image outputs, addressing token sequence alterations with a custom tokenizer finetuning and synchronization layer.

Findings

Enables reliable watermark detection with statistically grounded p-values.

Improves robustness against image transformations, compression, and removal attacks.

Demonstrates effectiveness through extensive experiments.

Abstract

Watermarking the outputs of generative models has emerged as a promising approach for tracking their provenance. Despite significant interest in autoregressive image generation models and their potential for misuse, no prior work has attempted to watermark their outputs at the token level. In this work, we present the first such approach by adapting language model watermarking techniques to this setting. We identify a key challenge: the lack of reverse cycle-consistency (RCC), wherein re-tokenizing generated image tokens significantly alters the token sequence, effectively erasing the watermark. To address this and to make our method robust to common image transformations, neural compression, and removal attacks, we introduce (i) a custom tokenizer-detokenizer finetuning procedure that improves RCC, and (ii) a complementary watermark synchronization layer. As our experiments demonstrate, our approach enables reliable and robust watermark detection with theoretically grounded p-values. Code and models are available at https://github.com/facebookresearch/wmar.