AOR-Bench: Do Large Audio Language Models Over-Refuse Pseudo-Harmful Queries?

2026-06-19

AI Summary

This study addresses the prevalent issue of over-rejection in current large audio language models when processing acoustically benign yet superficially harmful queries, which undermines the balance between usability and safety. The authors introduce AOR-Bench, the first benchmark specifically designed to evaluate over-rejection in audio language models, comprising 3,000 pseudo-harmful audio samples across six diverse scenarios. Through systematic evaluation of twelve state-of-the-art models, the work reveals the critical influence of acoustic context on safety judgments. To mitigate this issue, the paper proposes a lightweight strategy combining chain-of-thought reasoning with activation manipulation, which significantly reduces false rejection rates and demonstrates strong empirical effectiveness.

Abstract

Large Audio Language Models (LALMs) have demonstrated strong performance across a wide range of audio tasks. As they are increasingly deployed in real-world applications, ensuring their safety alignment has become more important. Although refusal mechanisms serve as a key safeguard by preventing LALMs from responding to harmful requests, they can also lead to {\em over-refusal}, where models incorrectly reject benign queries. This issue is especially challenging in the audio domain because speech that appears harmful in isolation may become benign when interpreted together with the surrounding acoustic context, such as background sounds. To study this problem, we introduce \textbf{AOR-Bench} (\textbf{A}udio \textbf{O}ver-\textbf{R}efusal \textbf{Bench}mark), the first benchmark for over-refusal specifically designed for LALMs. AOR-Bench contains 3,000 pseudo-harmful audio samples across six scenario categories. Evaluating 12 representative LALMs from six major model families, we find that over-refusal is widespread (Figure~\ref{fig:overall_performance}) and uncover several important patterns in their safety judgments. As a preliminary effort to mitigate this issue, we further explore two lightweight strategies (e.g., Chain-of-Thought and activation steering) to reduce over-refusal.

Problem

• over-refusal
• Large Audio Language Models
• safety alignment
• pseudo-harmful queries
• audio context

Innovation

• Audio Over-Refusal
• Large Audio Language Models
• Safety Alignment
• AOR-Bench
• Context-Aware Refusal