"I am bad": Interpreting Stealthy, Universal and Robust Audio Jailbreaks in Audio-Language Models
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:4RGVVSUBrecord.jsonopen to challenge →
read the original abstract
The rise of multimodal large language models has introduced innovative human-machine interaction paradigms but also significant challenges in machine learning safety. Audio-Language Models (ALMs) are especially relevant due to the intuitive nature of spoken communication, yet little is known about their failure modes. This paper explores audio jailbreaks targeting ALMs, focusing on their ability to bypass alignment mechanisms. We construct adversarial perturbations that generalize across prompts, tasks, and even base audio samples, demonstrating the first universal jailbreaks in the audio modality, and show that these remain effective in simulated real-world conditions. Beyond demonstrating attack feasibility, we analyze how ALMs interpret these audio adversarial examples and reveal them to encode imperceptible first-person toxic speech - suggesting that the most effective perturbations for eliciting toxic outputs specifically embed linguistic features within the audio signal. These results have important implications for understanding the interactions between different modalities in multimodal models, and offer actionable insights for enhancing defenses against adversarial audio attacks.
This paper has not been read by Pith yet.
Forward citations
Cited by 4 Pith papers
-
Benign Fine-Tuning Breaks Safety Alignment in Audio LLMs
Benign fine-tuning on audio data breaks safety alignment in Audio LLMs by raising jailbreak success rates up to 87%, with the dominant risk axis depending on model architecture and embedding proximity to harmful content.
-
Audio Jailbreaks in Large Audio-Language Models: Taxonomy, Attack-Defense Analysis, and Cost-Aware Evaluation
Organizes audio jailbreaks into semantic/acoustic/signal/embedding categories, evaluates representative attacks and defenses on ten LALMs with success rate plus latency and benign refusal, and concludes that acoustic ...
-
Towards Holistic Evaluation of Large Audio-Language Models: A Comprehensive Survey
The survey introduces a four-category taxonomy for LALM evaluations and reviews benchmarks across general auditory processing, knowledge reasoning, dialogue, and fairness-safety.
-
A Survey of Large Audio Language Models: Generalization, Trustworthiness, and Outlook
A survey of Large Audio Language Models that establishes a taxonomy of trustworthiness vulnerabilities and proposes a Defense-in-Depth roadmap for audio intelligence.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.