Purpose: This study examines the dual affective effects of Artificial Intelligence (AI)-powered personalized advertising on consumer trust and purchase intention in TikTok Shop, with Agreeableness and Neuroticism as moderators.  
Research Methodology: A quantitative survey of 355 active TikTok Shop users in Indonesia was conducted and analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) with SmartPLS 3.2.9.  
Results: AI Perceived Personalization enhances Perceived Emotional Support but does not trigger Emotional Creepiness. Perceived Emotional Support positively affects Trust, while Emotional Creepiness negatively affects Trust, and Trust drives Purchase Intention (PI). The positive path through Perceived Emotional Support significantly mediates the personalization–trust relationship, while the negative path does not. Agreeableness and Neuroticism significantly moderate affective responses.  
Conclusions: AI personalization mainly works through positive affect, particularly emotional support, to build trust and PI. Creepiness rarely arises but damages trust when it does. Affective responses are asymmetric, with positive reactions being more salient in the Indonesian TikTok Shop context.  
Limitations: The study is limited by a predominantly young, digitally intensive, personalization-exposed Indonesian sample, the absence of regional domicile data, reliance on purchase intention rather than actual behavior, and examination of only Agreeableness and Neuroticism.  
Contributions: Extends Affective Event Theory (AET) into a dual-affective-route asymmetric model, integrates Big Five personality traits as boundary conditions, and contextualizes AI personalization research in emerging markets.