Evaluating the Effectiveness of Institutional AI Ethics Policies in Promoting Academic Integrity: A Conceptual Framework

Abstract

The rapid adoption of generative artificial intelligence (AI) in higher education has transformed teaching, learning, and assessment practices, while intensifying concerns about academic integrity. Although many universities have introduced institutional AI ethics policies to promote responsible AI use, there remains limited evidence regarding the effectiveness of these policies in shaping students’ academic integrity behaviours. Existing studies predominantly focus on students’ ethical awareness and perceptions within single institutions and cross-sectional contexts, offering insufficient insight into the long-term behavioural impact of institutional governance mechanisms.
This paper develops a conceptual framework for evaluating the effectiveness of institutional AI ethics policies in promoting academic integrity in higher education. Drawing on the Theory of Planned Behaviour and Institutional Governance Theory, AI ethics policies are conceptualized as a multidimensional construct comprising policy clarity, AI ethics training, and enforcement mechanisms. Methodologically, the paper adopts a conceptual and analytical approach, integrating established theoretical perspectives with illustrative examples from Malaysian public and private universities, situated within national quality assurance expectations articulated by the Malaysian Qualifications Agency (MQA, 2023) and UNESCO (2023).
The proposed framework identifies how policy clarity and training enhance students’ ethical awareness and normative beliefs, while enforcement mechanisms strengthen perceived behavioural control and accountability. Collectively, these dimensions mediate the relationship between institutional AI ethics policies and academic integrity outcomes, including ethical AI use and reduced AI-related misconduct. Illustrative cases demonstrate that institutions implementing coordinated and consistently enforced policy components are better positioned to influence student behaviour meaningfully. By examining this gap, the framework aims to support higher education institutions in enhancing ethical AI use.