IntroductionAI adoption is growing rapidly across Chinese higher education, yet few studies explore local university stakeholders' perceptions of domestic generative AI (DeepSeek) for student administrative work. This paper fills this research gap.MethodsWe conducted a cross-sectional quantitative survey with 386 respondents (undergraduates, postgraduates, faculty, university administrative staff) at Panzhihua University. All data analysis was completed in R, testing five research hypotheses based on TAM and UTAUT frameworks. Bonferroni correction was applied to control Type I error for multiple comparisons. Composite scores are only used for descriptive analysis due to low internal consistency across measurement items.ResultsThe overall awareness rate of DeepSeek among participants reached 91.5%. Respondents rated administrative functions of AI (student information management M = 3.74, service delivery M = 3.72) far higher than mental health support functions (M = 3.17). The top three concerns were data privacy (M = 3.86), system transparency (M = 3.78), and ethical risks of AI usage (M = 3.72). Behavioural intention to use AI (M = 3.77) was higher than overall satisfaction (M = 3.56). Hypothesis H1 was supported, while H2, H3, H4 and H5 were rejected after correction.DiscussionThe results support a phased AI promotion strategy prioritizing administrative scenarios, with strict data governance and hybrid human-AI collaboration models for student mental health services. Key limitations include single-university non-probability sampling and weak reliability of composite measurement scales.