Purpose: This study examines how consumers evaluate AI-generated influencers in social commerce, focusing on credibility, perceived authenticity, trust, and AI awareness. It asks whether credibility alone builds trust in an artificial communicator, or whether authenticity is the more proximal basis for trust and purchase intention.
Design/methodology/approach: A cross-sectional survey of 283 social-media users aged 18 to 40 in Chongqing, China was analyzed using internal-consistency and factorability diagnostics, multiple regression, mediation, and moderation analysis.
Findings: Authenticity was the strongest predictor of trust (β = .830); credibility was not significant in the full trust model (β = −.013). Trust was the strongest predictor of purchase intention (β = .591); credibility (β = .191) and AI awareness (β = .172) also contributed. Simple mediation through trust was significant but disappeared once authenticity and AI awareness were controlled. The credibility × AI-awareness interaction was nonsignificant.
Research limitations/implications: The study relies on convenience sampling, self-report data, and two-item measures; future work should use multi-item scales and experimental or longitudinal designs.
Practical implications: AI influencer communication should prioritize verifiable information and transparent disclosure over human-like appearance, with stronger oversight in high-risk categories.
Social implications: Responsible practice should protect consumers from deceptive anthropomorphism so novelty is not mistaken for trustworthiness.
Originality/value: The study separates credibility, a source-quality cue, from authenticity, a more proximal judgment of genuineness, showing that a simple credibility–trust–purchase account overstates credibility’s role once authenticity is modeled simultaneously.