With the proliferation of personalized recommendation systems (PRS) such as Facebook, Amazon, and TikTok, the academic community has increasingly focused on AI-based recommendation systems. However, there remains a dearth of research that elucidates the adverse effects of AI-based recommendation algorithms and the underlying mechanisms through which these effects influence psychological and behavioral responses, particularly in terms of purchasing behavior. Drawing upon the Stressor-Strain-Outcome (SSO) model, this study aims to scrutinize the purported features of “greediness” and “bias” inherent in such algorithms, thereby investigating their impact on users’ negative psychological and behavioral responses. This investigation collected 473 online responses and used the partial least squares structural equation model (PLS-SEM) to empirically analyze the research model. The findings suggest that greedy and biased recommendation algorithms engender information narrowing, redundancy, overload, technological intrusiveness, and concerns about information disclosure. These stressors are associated with negative psychological and behavioral responses among users, which ultimately influence purchasing resistance behaviors on short video platforms. Consequently, the findings of this study contribute to a deeper understanding of the adverse implications of AI recommendation algorithms and provide valuable information for companies that offer short-form video applications.