This study empirically analyzes the impact of AI recommendation algorithm characteristics on ad fatigue and ad avoidance in short-form advertising. While short-form content platforms experiencing rapid growth in the digital media environment provide personalized advertisements through AI-based recommendation algorithms, this research originated from the recognition that such personalization may paradoxically induce ad fatigue and avoidance behaviors. The study established confirmation bias, perceived intrusiveness, perceived optimization, and AI algorithm trustworthiness as major independent variables, with ad fatigue as a mediating variable to examine their effects on ad avoidance. The results revealed that perceived intrusiveness had a strong positive effect on ad fatigue, while confirmation bias and AI algorithm trustworthiness had negative effects on ad fatigue. Perceived optimization did not significantly affect ad fatigue but had a direct negative effect on ad avoidance. Ad fatigue had a strong positive effect on ad avoidance. The mediation analysis confirmed that perceived intrusiveness and AI algorithm trustworthiness significantly influenced ad avoidance through ad fatigue as a mediator. This study represents the first systematic investigation of the negative effect mechanisms of AI recommendation algorithms in short-form advertising environments, contributing to the expansion of digital marketing theory. Practically, it provides algorithm optimization strategies to mitigate ad fatigue and avoidance.