[Purpose/significance] Existing studies have investigated factors affecting privacy risks in human-AI inter⁃actions, but their conclusions vary widely. This study systematically reviews relevant empirical research to identify key influencing factors and potential moderators of privacy risk perception in human-AI interaction. [Method/process] A meta-analysis was conducted, screening 64 valid studies. Key factors affecting privacy risk perception were identified across four dimensions: AI interaction characteristics, user characteristics, privacy concerns, and privacy protection en⁃vironments, totaling 19 critical factors. Potential moderators were tested to identify sources of heterogeneity in the find⁃ings. [Result/conclusion] Factors across the four dimensions influence users’privacy risk perception to varying de⁃grees. Information sensitivity, privacy concerns, perceived threats, and anxiety exhibit strong correlations with privacy risk perception. AI characteristics, long-term orientation, uncertainty avoidance, and gender demonstrate significant moderating effects.