pith. sign in

The devil behind the mask: An emergent safety vulnerability of diffusion LLMs

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

fields

cs.LG 1

years

2026 1

verdicts

UNVERDICTED 1

clear filters

representative citing papers

Backdooring Masked Diffusion Language Models

cs.LG · 2026-05-19 · unverdicted · novelty 7.0

SHADOWMASK backdoors MDLMs by replacing the all-mask terminal distribution with a trigger-mask mixture prior, achieving near-100% attack success on DiT and LLaDA-8B models across multiple datasets while resisting fine-tuning and some defenses.

citing papers explorer

Showing 0 of 0 citing papers after filters.

No citing papers match the current filters.