LR-DWM: Efficient Watermarking for Diffusion Language Models

TL;DR

This paper presents LR-DWM, an efficient watermarking scheme for diffusion language models that embeds detectable signals with minimal computational overhead, enabling reliable attribution of AI-generated text.

Contribution

LR-DWM introduces a novel watermarking method for diffusion language models that biases tokens based on both neighbors, reducing overhead while maintaining detectability.

Findings

High detectability of watermarks in DLMs

Minimal runtime and memory overhead

Effective watermarking close to non-watermarked models

Abstract

Watermarking (WM) is a critical mechanism for detecting and attributing AI-generated content. Current WM methods for Large Language Models (LLMs) are predominantly tailored for autoregressive (AR) models: They rely on tokens being generated sequentially, and embed stable signals within the generated sequence based on the previously sampled text. Diffusion Language Models (DLMs) generate text via non-sequential iterative denoising, which requires significant modification to use WM methods designed for AR models. Recent work proposed to watermark DLMs by inverting the process when needed, but suffers significant computational or memory overhead. We introduce Left-Right Diffusion Watermarking (LR-DWM), a scheme that biases the generated token based on both left and right neighbors, when they are available. LR-DWM incurs minimal runtime and memory overhead, remaining close to the non-watermarked baseline DLM while enabling reliable statistical detection under standard evaluation settings. Our results demonstrate that DLMs can be watermarked efficiently, achieving high detectability with negligible computational and memory overhead.