Trustworthy AI in education: Framework, cases, and governance strategies

Artificial intelligence (AI) in education has triggered significant debates about ethics, fairness, and accountability. This article introduces a five-dimensional trust framework to evaluate educational AI systems across five key domains: privacy, safety, fairness, explainability, and accountability. Through a comparative analysis of four representative cases—a virtual teaching assistant, an adaptive learning platform, an algorithmic grading system, and an AI-based proctoring tool—we identify recurring trust-related risks and systemic governance vulnerabilities, such as insufficient oversight, lack of transparency, and unclear accountability mechanisms. The analysis reveals risks such as data breaches, bias, opaque decision-making, and ambiguous responsibility. To address these issues, we propose multi-stakeholder governance strategies, including ethical-by-design principles, institutional oversight, and regulatory standards. This study concludes by outlining major avenues for future research, such as developing explainable AI systems tailored to educational settings, constructing trust models for human–AI collaboration, and evaluating the enduring impacts of AI governance structures.