SafeSearch: Automated Red-Teaming of LLM-Based Search Agents

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Dong, Jianshuo, Guo, Sheng, Wang, Hao, Chen, Xun, Liu, Zhuotao, Zhang, Tianwei, Xu, Ke, Huang, Minlie, Qiu, Han
Format:	Preprint
Published:	2025
Subjects:	Artificial Intelligence Computation and Language Cryptography and Security
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

_version_	1866918529261895680
author	Dong, Jianshuo Guo, Sheng Wang, Hao Chen, Xun Liu, Zhuotao Zhang, Tianwei Xu, Ke Huang, Minlie Qiu, Han
author_facet	Dong, Jianshuo Guo, Sheng Wang, Hao Chen, Xun Liu, Zhuotao Zhang, Tianwei Xu, Ke Huang, Minlie Qiu, Han
contents	Search agents connect LLMs to the Internet, enabling them to access broader and more up-to-date information. However, this also introduces a new threat surface: unreliable search results can mislead agents into producing unsafe outputs. Real-world incidents and our two in-the-wild observations show that such failures can occur in practice. To study this threat systematically, we propose SafeSearch, an automated red-teaming framework that is scalable, cost-efficient, and lightweight, enabling sandboxed safety evaluation of search agents. Using this, we generate 300 test cases spanning five risk categories (e.g., misinformation and prompt injection) and evaluate three search agent scaffolds across 17 representative LLMs. Our results reveal substantial vulnerabilities in LLM-based search agents, with the highest ASR reaching 90.5% for GPT-4.1-mini in a search-workflow setting. Moreover, we find that common defenses, such as reminder prompting, offer limited protection. Overall, SafeSearch provides a practical way to measure and improve the safety of LLM-based search agents.
format	Preprint
id	arxiv_https___arxiv_org_abs_2509_23694
institution	arXiv
publishDate	2025
record_format	arxiv
spellingShingle	SafeSearch: Automated Red-Teaming of LLM-Based Search Agents Dong, Jianshuo Guo, Sheng Wang, Hao Chen, Xun Liu, Zhuotao Zhang, Tianwei Xu, Ke Huang, Minlie Qiu, Han Artificial Intelligence Computation and Language Cryptography and Security Search agents connect LLMs to the Internet, enabling them to access broader and more up-to-date information. However, this also introduces a new threat surface: unreliable search results can mislead agents into producing unsafe outputs. Real-world incidents and our two in-the-wild observations show that such failures can occur in practice. To study this threat systematically, we propose SafeSearch, an automated red-teaming framework that is scalable, cost-efficient, and lightweight, enabling sandboxed safety evaluation of search agents. Using this, we generate 300 test cases spanning five risk categories (e.g., misinformation and prompt injection) and evaluate three search agent scaffolds across 17 representative LLMs. Our results reveal substantial vulnerabilities in LLM-based search agents, with the highest ASR reaching 90.5% for GPT-4.1-mini in a search-workflow setting. Moreover, we find that common defenses, such as reminder prompting, offer limited protection. Overall, SafeSearch provides a practical way to measure and improve the safety of LLM-based search agents.
title	SafeSearch: Automated Red-Teaming of LLM-Based Search Agents
topic	Artificial Intelligence Computation and Language Cryptography and Security
url	https://arxiv.org/abs/2509.23694