Backdooring Masked Diffusion Language Models

Abstract

Masked diffusion language models (MDLMs) are emerging as a compelling new paradigm for text generation, but their training-time security remains largely unexplored. Existing backdoor attacks on Gaussian diffusion models or autoregressive language models do not directly apply to MDLMs because MDLMs rely on discrete state corruption and iterative denoising rather than continuous noising or left-to-right prediction. In this work, we present the first systematic study of training-time backdoor attac

Categories

Framework mappings

Suggested from the entry's categories.

@misc{cao2026backdooring,
  title = {{Backdooring Masked Diffusion Language Models}},
  author = {Daniel Yiming Cao and Chengzhong Wang and Sheng-Yen Chou and Chengyu Huang and Pin-Yu Chen and Shengwei An},
  year = {2026},
  month = may,
  eprint = {2605.19262},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.19262}
}