Triggered to buy? The role of negative emotions and user awareness of others in driving impulse purchases in AI live streaming

Purpose Based on mood management theory and social facilitation theory, this study examines how anthropomorphism and automated social presence of AI live streamers influence viewers’ emotions and impulse buying behavior, and how the awareness of other viewers moderates these impacts. Design/methodology/approach A cross-sectional questionnaire survey was used to collect 528 responses from the Greater Bay Area. AI-hosted live streaming video selling underwear was shown first, and then respondents completed an online survey. The relationships among constructs in the theoretical framework, measurement and structural models were examined using Partial Least Squares Structural Equation Modeling and PLSpredict. Findings Anthropomorphism and automated social presence were important antecedents of emotions, except for the nonsignificant impact of automated social presence on negative emotions. Positive emotions had a strong positive effect on impulse buying, whereas negative emotions had a weaker negative effect. Awareness of other viewers significantly moderated the negative emotion relations. Under high social visibility, negative emotions no longer inhibited impulse buying and instead increased it. Originality/value While impulse buying behavior is common in a live streaming, little research has examined how social presence and emotions influence viewers’ impulse buying decisions. This study extends mood management and social facilitation theories by demonstrating that emotional effects in AI live streaming depend on social cues rather than purely individual factors. The findings offer practical guidance for brands on how AI avatars, emotions and social visibility can be strategically managed to influence consumer decision-making in real-time live commerce.