Sparse Tokens Suffice: Jailbreaking Audio Language Models via Token-Aware Gradient Optimization 事件

Sparse Tokens Suffice: Jailbreaking Audio Language Models via Token-Aware Gradient Optimization arXiv:2605.04700v2 Announce Type: replace-cross Abstract: Jailbreak attacks on audio language models (ALMs) optimize audio perturbations to elicit unsafe generations, and they typically update the entire waveform densely throughout optimization. In this work, we investigate the necessity of such dense optimization by analyzing the structure of token-aligned gradients in ALMs. We find that gradient en