From Acoustic Sensing to Physical Actions: A Comprehensive Survey of Security Risks in Edge-Deployed Audio LLM-Based Agents
cs.CR, cs.CV, cs.LG
Submitted: 2026-10-11
Updated: 2026-10-11
Sources
- VoxPrivacy: A Benchmark for Evaluating Interactional Privacy of Speech Language Models
- Aegis: Towards Governance, Integrity, and Security of AI Voice Agents
- Security and Privacy Problems in Voice Assistant Applications: A Survey
- Towards Holistic Evaluation of Large Audio-Language Models: A Comprehensive Survey
- A Survey of Large Audio Language Models: Generalization, Trustworthiness, and Outlook
- Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation
- A Survey on Long-Term Memory Security in LLM Agents: Attacks, Defenses, and Governance Across the Memory Lifecycle
- A Review on Edge Large Language Models: Design, Execution, and Applications
- SALMONN: Towards Generic Hearing Abilities for Large Language Models
- Qwen-Audio: Advancing Universal Audio Understanding via Unified Large-Scale Audio-Language Models
- AudioPaLM: A Large Language Model That Can Speak and Listen
- Moshi: a speech-text foundation model for real-time dialogue
- Adaptive Edge-Cloud Inference for Speech-to-Action Systems Using ASR and Large Language Models
- Bridging the Perception Gap: A Lightweight Coarse-to-Fine Architecture for Edge Audio Systems
- Sirens' Whisper: Inaudible Near-Ultrasonic Jailbreaks of Speech-Driven LLMs
- Hijacking Large Audio-Language Models via Context-Agnostic and Imperceptible Auditory Prompt Injection
- Abusing Images and Sounds for Indirect Instruction Injection in Multi-Modal LLMs
- Voice Jailbreak Attacks Against GPT-4o
- On Optimizing Multimodal Jailbreaks for Spoken Language Models
- Mind the GAP: Text Safety Does Not Transfer to Tool-Call Safety in LLM Agents
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs