arXiv AI August 20, 2026 By Linghan Huang, Bo Li, Huaming Chen, Kim-Kwang Raymond Choo

`From Prompt to Perturbation': An Adaptive Framework for Voice-Based Jailbreaks on Audio LLMs

The paper introduces an adaptive jailbreak attack framework that evaluates both cascaded pipelines and end‑to‑end large audio‑language models (LALMs) under a unified setting. It employs a feedback‑guided mutation engine to automatically generate and refine jailbreak candidates across textual prompts and audio perturbations, thereby broadening attack diversity. Experiments on six audio‑based systems show that both paradigms remain highly vulnerable, with the framework achieving higher attack success rates than existing methods.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

arXiv Machine Learning

Jul 1

On Optimizing Multimodal Jailbreaks for Spoken Language Models

arXiv:2603. 19127v2 Announce Type: replace Abstract: As Spoken Language Models (SLMs) integrate speech and text modalities, they inherit the safety vulnerabilities of their LLM backbone while introducing an expanded attack surface.

By Aravind Krishnan, Karolina Sta\'nczak, Dietrich Klakow

llms multimodal benchmarks safety

arXiv AI

Jun 2

The Alignment Curse: Modality Alignment Supercharges Audio Attacks via Text Transfer

arXiv:2602. 02557v2 Announce Type: replace-cross Abstract: Recent advances in end-to-end trained omni-models have substantially improved audio capabilities by strengthening text-audio modality alignment.

By Yupeng Chen, Junchi Yu, Aoxi Liu, Baoyuan Wu, Philip Torr, Adel Bibi

safety

arXiv AI

Sep 12

Spectral Masking and Interpolation Attack (SMIA): A Black-box Adversarial Attack against Voice Authentication and Anti-Spoofing Systems

The paper introduces the Spectral Masking and Interpolation Attack (SMIA), a black‑box adversarial technique that subtly alters inaudible frequency regions of AI‑generated audio to fool voice authentication systems and their countermeasures. Experiments show SMIA achieves at least 82% success against combined verification and countermeasure systems, 97.5% against standalone speaker verification, and 100% against countermeasures, revealing a critical security gap. The authors argue that current static defenses are inadequate and call for dynamic, context‑aware defenses that can adapt to evolving threats.

By Kamel Kamel, Hridoy Sankar Dutta, Keshav Sood, Sunil Aryal

multimodal benchmarks safety

arXiv AI

Aug 10

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

arXiv:2604. 09222v2 Announce Type: replace-cross Abstract: Audio Large Language Models (ALLMs) enable spoken interaction but introduce new jailbreak vulnerabilities.

By Yunqiang Wang, Hengyuan Na, Di Wu, Miao Hu, Guocong Quan

llms safety

arXiv Computation and Language

Sep 23

Dynamic Deep Prompt Optimization for Defending Against Jailbreak Attacks on LLMs

arXiv:2609.26185v1 Announce Type: cross Abstract: Large Language Models (LLMs) demonstrate impressive capabilities across many applications but remain vulnerable to jailbreak attacks, which elicit ha...

By Doniyorkhon Obidov, Honggang Yu, Xiaolong Guo, Kaichen Yang

llms rag fine-tuning safety

arXiv AI

Oct 15

Exploring Jailbreak Vulnerability Across Depth-Based Dynamic Inference in Large Language Models

arXiv:2610.15207v1 Announce Type: cross Abstract: Large Language Models (LLMs) have been shown to be computationally expensive and vulnerable to jailbreak attacks, in which adversarial prompts bypass...

By Lass Najmuldeen, Seyedhamidreza Mousavi

llms safety

One email a morning, machine-written

Related stories

On Optimizing Multimodal Jailbreaks for Spoken Language Models

The Alignment Curse: Modality Alignment Supercharges Audio Attacks via Text Transfer

Spectral Masking and Interpolation Attack (SMIA): A Black-box Adversarial Attack against Voice Authentication and Anti-Spoofing Systems

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

Dynamic Deep Prompt Optimization for Defending Against Jailbreak Attacks on LLMs

Exploring Jailbreak Vulnerability Across Depth-Based Dynamic Inference in Large Language Models