Ethics of AI in Education

Generative and predictive AI systems are rapidly moving from the margins of the education system, such as optional tutoring tools and reference aids, into the core of instructional design, including automated feedback, adaptive learning, proctoring, and AI-mediated student support. This shift introduces a clear concern: while educational stakeholders value the scale, personalization, and efficiency promised by AI, these systems can also introduce ethical risks that are difficult to observe within classrooms, such as privacy leakage, disparate impact, opaque decision-making, erosion of academic integrity, and overreliance on uncertain AI-generated outputs. Although prior scholarship and policy frameworks provide general principles, educators and administrators still lack operational guidance that connects governance choices, such as disclosure practices and policy enactment and enforcement, to student-centered outcomes, including trust, perceived fairness, and learning engagement. To address this gap, we contribute both an actionable standards-to-controls matrix tailored to AI-in-education frameworks and an illustrative, fully specified field study in a business school context, demonstrating how governance features can be evaluated quantitatively. Using a three-condition experimental design, no explicit AI policy, AI policy enactment, and AI policy enactment with explicit enforcement, our results show significant differences in perceived transparency and trust in education. We conclude by discussing implications for policy design, academic instructional practice, and future research on auditable, education-aligned AI governance.