The Hidden Curriculum of Generative AI: Institutional AI Policies and the Potential for Student Inequality

Abstract

Generative artificial intelligence (GenAI) is increasingly integrated into higher education, prompting institutions to develop policies governing its use in teaching, learning, and assessment. Although existing discussions have emphasized academic integrity, assessment practices, and technological access, less attention has been given to the knowledge and competencies students may need to interpret and comply with institutional AI policies. This study examines how institutional GenAI policies establish explicit and implicit expectations for students and identifies potential pathways through which policy-navigation demands may contribute to educational inequality. Using comparative qualitative content analysis, publicly available policies and guidance documents from 20 purposively selected higher education institutions across the United States, Canada, United Kingdom, Australia, and South Africa were examined. The analysis considered policy permissiveness, disclosure requirements, instructor discretion, student responsibilities, AI-literacy guidance, accessibility, equity, and implicit navigation demands. Findings indicate that acceptable AI use is frequently determined at the course, instructor, or assessment level, requiring students to interpret context-specific rules, understand disclosure requirements, verify AI-generated outputs, and reconcile varying instructor expectations. However, explicit consideration of differences in technological access, language, accessibility, and institutional knowledge was less consistently evident across the reviewed documents. Drawing on hidden-curriculum and digital-inequality perspectives, the study identifies AI policy-navigation burdens and proposes the concept of a potential policy-navigation divide. While the findings do not establish unequal student experiences or educational outcomes, they highlight institutional conditions through which inequalities could emerge. The study recommends clearer policy communication, transparent assessment-specific expectations, accessible AI-literacy support, and greater attention to equity in institutional AI governance.