AuditAgent: LLM Agent for Risks Auditing in Recommender Systems

Auditing recommendation systems has attracted growing attention due to increasing concerns over filter bubbles, unfairness, and data misuse. A common approach is sock-puppet auditing, where autonomous agents interact with platforms to reveal risks. However, existing approaches rely on hard-coded agents, lacking adaptability to dynamic GUI layouts and generating behaviors far from those of real users, limiting the comprehensiveness and representativeness of assessment. To address these issues, we introduce AuditAgent, an LLM-powered GUI-agent framework for risk auditing. AuditAgent simulates realistic user preferences and performs adaptive, human-like interactions on recommendation platforms. This design enables more thorough and faithful auditing, providing comprehensive assessments across multiple risk dimensions, including filter bubbles, unfairness, and data misuse.

Authors

• Du Su State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences
• Zhenxing Chen University of Chinese Academy of Sciences
• Shilong Zhao State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences University of Chinese Academy of Sciences
• Yuanhao Liu State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences University of Chinese Academy of Sciences
• Fei Sun CAS Key Laboratory of AI Safety & Security, Institute of Computing Technology, Chinese Academy of Sciences
• Qi Cao
• Huawei Shen CAS Key Laboratory of AI Safety & Security, Institute of Computing Technology, Chinese Academy of Sciences University of Chinese Academy of Sciences

Keywords

No keywords are indexed for this paper.