Saved in:
Bibliographic Details
Main Authors: Zheng, Xiang, Wang, Longxiang, Liu, Yi, Ma, Xingjun, Shen, Chao, Wang, Cong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.02997
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913637097013248
author Zheng, Xiang
Wang, Longxiang
Liu, Yi
Ma, Xingjun
Shen, Chao
Wang, Cong
author_facet Zheng, Xiang
Wang, Longxiang
Liu, Yi
Ma, Xingjun
Shen, Chao
Wang, Cong
contents Auditing Large Language Models (LLMs) is a crucial and challenging task. In this study, we focus on auditing black-box LLMs without access to their parameters, only to the provided service. We treat this type of auditing as a black-box optimization problem where the goal is to automatically uncover input-output pairs of the target LLMs that exhibit illegal, immoral, or unsafe behaviors. For instance, we may seek a non-toxic input that the target LLM responds to with a toxic output or an input that induces the hallucinative response from the target LLM containing politically sensitive individuals. This black-box optimization is challenging due to the scarcity of feasible points, the discrete nature of the prompt space, and the large search space. To address these challenges, we propose Curiosity-Driven Auditing for Large Language Models (CALM), which uses intrinsically motivated reinforcement learning to finetune an LLM as the auditor agent to uncover potential harmful and biased input-output pairs of the target LLM. CALM successfully identifies derogatory completions involving celebrities and uncovers inputs that elicit specific names under the black-box setting. This work offers a promising direction for auditing black-box LLMs. Our code is available at https://github.com/x-zheng16/CALM.git.
format Preprint
id arxiv_https___arxiv_org_abs_2501_02997
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CALM: Curiosity-Driven Auditing for Large Language Models
Zheng, Xiang
Wang, Longxiang
Liu, Yi
Ma, Xingjun
Shen, Chao
Wang, Cong
Artificial Intelligence
Computation and Language
Auditing Large Language Models (LLMs) is a crucial and challenging task. In this study, we focus on auditing black-box LLMs without access to their parameters, only to the provided service. We treat this type of auditing as a black-box optimization problem where the goal is to automatically uncover input-output pairs of the target LLMs that exhibit illegal, immoral, or unsafe behaviors. For instance, we may seek a non-toxic input that the target LLM responds to with a toxic output or an input that induces the hallucinative response from the target LLM containing politically sensitive individuals. This black-box optimization is challenging due to the scarcity of feasible points, the discrete nature of the prompt space, and the large search space. To address these challenges, we propose Curiosity-Driven Auditing for Large Language Models (CALM), which uses intrinsically motivated reinforcement learning to finetune an LLM as the auditor agent to uncover potential harmful and biased input-output pairs of the target LLM. CALM successfully identifies derogatory completions involving celebrities and uncovers inputs that elicit specific names under the black-box setting. This work offers a promising direction for auditing black-box LLMs. Our code is available at https://github.com/x-zheng16/CALM.git.
title CALM: Curiosity-Driven Auditing for Large Language Models
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2501.02997