1
Codec-Robust Attacks on Audio LLMs
针对音频大模型的攻击新范式，能在编解码预处理下依然生效，突破现有防御机制。
arXiv:2605.20519v1 Announce Type: cross Abstract: Prior attacks on Audio Large Language Models (Audio LLMs) demonstrated that carefully crafted wavefo…
针对音频大模型的攻击新范式，能在编解码预处理下依然生效，突破现有防御机制。
arXiv:2605.20519v1 Announce Type: cross Abstract: Prior attacks on Audio Large Language Models (Audio LLMs) demonstrated that carefully crafted wavefo…