GRM: Utility-Aware Jailbreak Attacks on Audio LLMs via Gradient-Ratio Masking

Abstract

Audio large language models (ALLMs) enable rich speech-text interaction, but they also introduce jailbreak vulnerabilities in the audio modality. Existing audio jailbreak methods mainly optimize jailbreak success while overlooking utility preservation, as reflected in transcription quality and question answering performance. In practice, stronger attacks often come at the cost of degraded utility. To study this trade-off, we revisit existing attacks by varying their perturbation coverage in the

Categories

Framework mappings

Suggested from the entry's categories.

@misc{wang2026grm,
  title = {{GRM: Utility-Aware Jailbreak Attacks on Audio LLMs via Gradient-Ratio Masking}},
  author = {Yunqiang Wang and Hengyuan Na and Di Wu and Miao Hu and Guocong Quan},
  year = {2026},
  month = apr,
  eprint = {2604.09222},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2604.09222}
}