<p>With the application of artificial intelligence in physical education, how physical education teachers understand and embrace AI technology in their actual teaching has become a key issue in the reform of physical education. However, existing research has paid insufficient attention to physical education teachers’ cognitive states and their influencing factors, making it difficult to provide a basis for effectively integrating AI technology into physical education classrooms. This study aims to reveal physical education teachers’ perceptions of AI and clarify the core factors influencing their acceptance of AI technology. Based on Social Cognitive Theory, the research primarily sets three objectives: First, to analyse physical education teachers’ basic perceptions and attitudes toward AI; second, to examine the personal, environmental, and behavioural factors affecting their acceptance of AI technology; and third, to construct a model of the mechanism underlying physical education teachers’ acceptance of AI technology. This study employed qualitative research methods, utilising purposive sampling to conduct semi-structured interviews with 16 physical education teachers. Data coding and thematic analysis were performed using NVivo software. Findings revealed that a combination of multidimensional factors, including personal, environmental, situational, and resource factors, influences physical education teachers’ acceptance of AI. Among these, self-efficacy serves as the core factor driving physical education teachers’ acceptance of AI. Outcome expectancy significantly influences behavioural decisions, while subjective norms mediate the relationship between self-efficacy and these decisions. By integrating SCT with SDT, this study provides a more context-sensitive explanation of PE teachers’ AI acceptance in embodied teaching settings.</p>