Responsible AI Governance in Iran: A Program-Theory Evaluation of the National Artificial Intelligence Strategy

Abstract

Artificial intelligence has become a strategic technology whose development requires coordinated institutional, regulatory, ethical, and infrastructural governance. This study evaluated the capacity of Iran’s National Artificial Intelligence Strategy to support responsible AI governance by reconstructing its underlying program theory and examining the coherence of its implementation architecture. The research adopted a qualitative, theory-based case-study design. Data were obtained through documentary analysis of the National Artificial Intelligence Strategy and related policy materials, together with semi-structured interviews with 12 experts from academic, research, policy, and industrial backgrounds. The data were coded and thematically analysed using MAXQDA, after which an integrated theory of change and theory of action was developed and applied to the national strategy. The findings indicated that responsible AI governance in Iran depends on sustained political commitment, a legally empowered and agile coordinating institution, equitable access to high-quality data and computational infrastructure, interdisciplinary human capital, effective university–industry collaboration, risk-based regulation, and continuous policy evaluation. The strategy provides an important normative foundation by emphasizing human dignity, national values, scientific development, priority sectors, and the participation of knowledge-based firms. Nevertheless, its implementation architecture remains insufficiently specified. Major weaknesses include ambiguity regarding institutional authority, sustainable financing, data classification, access to computational resources, legal liability, regulatory instruments, measurable indicators, and mechanisms for periodic revision. The analysis also identified major policy dualities involving open versus sensitive data, centralized versus networked governance, strict versus adaptive regulation, domestic versus international standards, state versus private investment, and problem-oriented versus technology-oriented development. The study concludes that the effectiveness of the strategy depends on translating broad objectives into an explicit, funded, accountable, and adaptive program of action. Responsible AI governance requires not merely ethical declarations, but a coherent causal chain connecting institutions, resources, activities, outputs, societal outcomes, and mechanisms of accountability.