AI hallucinations and user verification intentions: an integrative model based on the DeLone and McLean model and the theory of planned behavior

Abstract

Generative artificial intelligence (GenAI) has become deeply integrated into everyday life and has increasingly become a key source of information. However, the prevalence of AI hallucinations, which refer to instances in which AI systems generate inaccurate or misleading information, raises concerns about the quality and reliability of the responses. Drawing on the Information Systems Success Model and the Theory of Planned Behavior (TPB), this study examines the factors that influence users' intentions to verify information generated by GenAI. An online survey was conducted among 620 Chinese participants. The findings based on structural equation modeling analysis reveal that system quality, information quality, and service quality positively influence perceived credibility, which in turn reduces perceptions of AI hallucination potential. Perceived hallucination potential further promotes a favorable attitude toward information verification. Consistent with the TPB, verification intention is significantly affected by attitude, subjective norm, and perceived behavioral control. This study extends the literature on AI hallucinations and information verification by identifying the mechanisms through which quality perceptions and credibility shape users' verification intentions. Theoretical implications and practical recommendations are discussed.