From Inaudible Inputs to Model Failures: Low-Frequency Safety Risks in LALMs

Hugging Face Daily Papers Papers

Summary

A paper introducing ILL, an inaudible low-frequency red-teaming method to expose safety vulnerabilities in audio-language models, and DRG, a defense that detects distribution shifts and requests a second recording to recover accuracy.

Large audio-language models (LALMs) have demonstrated strong capabilities in understanding diverse audio inputs. This diversity includes low-frequency signals that are inaudible to humans but can still enter the model and influence its generation. However, the practical impact of such low-frequency inputs on LALMs remains largely unexplored. In this paper, we propose Intermittent Low-Frequency Lockout (ILL), an inaudible red teaming method that evaluates this risk using a universal waveform template in a black box setting. ILL uses Sentence Attention Scale Estimation to determine active intervals and Frequency Confusion Transfer to construct a low-frequency waveform with continuous phase from corpus spectral variation. To mitigate this risk, we propose Distributional Requery Guard (DRG) to detect low-frequency distribution shifts and conditionally request a second recording for semantic recovery. Across six LALMs and multiple audio understanding tasks, ILL reduces accuracy by up to 67 percentage points while receiving a mean human audibility rating of 1.33, close to 1.17 for clean audio; DRG raises mean attacked accuracy from 28.5\% to 46.1\% after clean reacquisition. These findings identify a previously overlooked safety risk for LALMs and provide a foundation for future research on robust audio understanding.
Original Article
View Cached Full Text

Cached at: 08/14/26, 03:29 PM

Paper page - From Inaudible Inputs to Model Failures: Low-Frequency Safety Risks in LALMs

Source: https://huggingface.co/papers/2608.09158

Abstract

Researchers propose a black-box red-teaming method using inaudible low-frequency waveforms to expose vulnerabilities in audio-language models, alongside a defense that detects distribution shifts and requests a second recording to recover accuracy.

Large audio-language models(LALMs) have demonstrated strong capabilities in understanding diverse audio inputs. This diversity includes low-frequency signals that are inaudible to humans but can still enter the model and influence its generation. However, the practical impact of such low-frequency inputs on LALMs remains largely unexplored. In this paper, we proposeIntermittent Low-Frequency Lockout(ILL), an inaudible red teaming method that evaluates this risk using a universal waveform template in a black box setting. ILL usesSentence Attention Scale Estimationto determine active intervals andFrequency Confusion Transferto construct a low-frequency waveform with continuous phase from corpus spectral variation. To mitigate this risk, we proposeDistributional Requery Guard(DRG) to detect low-frequency distribution shifts and conditionally request a second recording for semantic recovery. Across six LALMs and multiple audio understanding tasks, ILL reduces accuracy by up to 67 percentage points while receiving a mean human audibility rating of 1.33, close to 1.17 for clean audio; DRG raises mean attacked accuracy from 28.5\% to 46.1\% after clean reacquisition. These findings identify a previously overlooked safety risk for LALMs and provide a foundation for future research on robust audio understanding.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2608\.09158

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.09158 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.09158 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.09158 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Voice AI Systems Are Vulnerable to Hidden Audio Attacks

Hacker News Top

New research shows that imperceptible audio signals can hijack large audio-language models (LALMs) with 79-96% success, forcing them to execute unauthorized commands like web searches or sending emails. The technique, dubbed AudioHijack, targets generative models and works regardless of user input, posing a serious security risk to voice AI systems.