Governing Generative AI in Higher Education: A Cross-National Analysis of Policy Framing and Student Maneuvering

Universities promote generative artificial intelligence (GenAI) as a pedagogical resource while regulating it as a risk to academic integrity and fair assessment. This study examines how differentiated policy framing varies across substantive policy dimensions and national contexts, and how students interpret and navigate such guidance in practice. Using a convergent mixed-methods design with quantitative priority, we analyzed GenAI policy documents from 36 universities in Denmark, the United Kingdom, the United States, and the Netherlands, complemented by eight semi-structured interviews with graduate students at one sampled university. Aspect-based sentiment analysis and mixed-effects modeling show significantly positive sentiment in teaching-related statements and significantly negative sentiment in GenAI- and ethics-related statements, while student- and assessment-related statements do not differ significantly from neutral. Neither country nor the policy-dimension-by-country interaction is statistically significant. The interviews contextualize these policy-level patterns: participants described uncertainty around acceptable use, disclosure, and instructor expectations, with responses ranging from cautious experimentation to avoidance. Together, the findings show how differentiated GenAI guidance can coexist within university policy and illustrate the interpretive demands students may face when applying such guidance in practice. Clearer programme- and course-level guidance, consistent disclosure expectations, and stronger support for GenAI literacy may help connect institutional policy more closely with teaching and learning practice.