Students’ intentions to truthfully declare generative AI use in higher education assessment: a cross-sectional SEM study integrating deterrence theory and the theory of planned behavior

Introduction As generative artificial intelligence (GenAI) becomes increasingly embedded in higher education, truthful declaration of GenAI use has become increasingly important for maintaining transparency and supporting the validity and integrity of academic assessment. Drawing on the Theory of Planned behavior (TPB) and deterrence theory, this study examines the psychological factors associated with students’ intentions to truthfully declare their GenAI use. The proposed framework distinguishes between deterrence targeting non-disclosure and deterrence associated with improper GenAI use exposed through truthful declaration. Methods Data were collected through the Credamo platform from currently enrolled university students in China who had experience with GenAI-use declaration in assessed work. A total of 466 valid questionnaires were obtained. Structural equation modeling (SEM) was employed to test the hypothesized relationships, and 5,000 bootstrap resamples were used to estimate confidence intervals for indirect associations. Results Attitude (β = 0.564), subjective norm (β = 0.219), and perceived behavioral control (β = 0.186) were positively associated with truthful declaration intention. Non-disclosure detection certainty (NDC) was positively associated with attitude (β = 0.643) and perceived behavioral control (β = 0.588), whereas improper-use identification certainty (IIC) was negatively associated with both constructs (β = −2.175 and −2.154, respectively). Non-disclosure sanction severity was negatively associated with attitude and perceived behavioral control (β = −0.486 and −0.690, respectively), whereas improper-use sanction severity was positively associated with both constructs (β = 1.916 and 1.892, respectively). All these structural paths were statistically significant ( p < 0.001). Bootstrap analyses identified significant indirect associations with intention via both attitude and perceived behavioral control: positive for non-disclosure detection certainty and improper-use sanction severity, and negative for non-disclosure sanction severity and improper-use identification certainty. Discussion The findings highlight the value of distinguishing deterrence targets and separating detection certainty from sanction severity. These perceptions showed contrasting associations with attitude and perceived behavioral control, with corresponding model-implied indirect associations with truthful declaration intention. The results inform differentiated approaches to disclosure governance, although the conditional sanction-severity estimates require cautious interpretation.