MaskForge: Structure-Aware Adaptive Attacks for Jailbreaking Diffusion Large Language Models

Abstract

Diffusion large language models (dLLMs) generate text by iteratively denoising partially masked sequences under bidirectional context, exposing a safety surface distinct from autoregressive LLMs. Because mask tokens are native inputs and tokens are committed by confidence rather than position, harmful content can be induced through infilling and outside the monitored prefix. Existing jailbreaks either miss this native infill capability or rely on low-diversity mask-bearing templates applied unif

Categories

Framework mappings

Suggested from the entry's categories.

@misc{ma2026maskforge,
  title = {{MaskForge: Structure-Aware Adaptive Attacks for Jailbreaking Diffusion Large Language Models}},
  author = {Yingzi Ma and Zhengyue Zhao and Xiaogeng Liu and Minhui Xue and Yue Zhao and Chaowei Xiao},
  year = {2026},
  month = jun,
  eprint = {2606.04027},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2606.04027}
}