Codec-Robust Attacks on Audio LLMs

ArXi:2605.20519v1 Announce Type: cross Prior attacks on Audio Large Language Models (Audio LLMs) nstrated that carefully crafted waveform-domain perturbations can force targeted adversarial outputs. As a defense mechanism against these attacks, real-world codec compression preprocessing has been studied to both detect and remove the perturbations. Yet no existing attack has nstrated robustness against these compressions. We