purpose: this study compares different models of artificial intelligence (ai) governance in order to propose an integrated framework adaptable to the conditions of countries and organizations.
methodology: the research is based on a systematic literature review، including the collection and screening of academic publications and policy reports، followed by a critical analysis of existing models. the sources are organized into two categories: (1) governance mechanism&-based models (centralized، decentralized، hybrid، and public&-private partnership models) and (2) principle- and goal-based models (rule-based، ethics-based، accountability، transparency، risk management، and multi‑stakeholder participation).
findings: no single model of ai governance is sufficient on its own. centralized models provide greater coordination but tend to be slower and more power‑concentrated. in contrast، decentralized models promote innovation and local adaptability but often face fragmented standards. hybrid models attempt to balance these approaches، while principle‑based models emphasize values such as fairness، transparency، and accountability; however، the connection between these values and binding implementation mechanisms often remains weak.
originality/value: this article proposes a two‑layer model for ai governance consisting of an operational layer (policy‑making، clear regulations، supervisory institutions، and accountability mechanisms) and an ethical layer (fairness، transparency، accountability، and data protection). these two layers are interconnected through shared institutions، common guidelines، and coordinated education. overall، the proposed framework aims to create a dynamic، evidence‑based governance structure that strengthens public trust، reduces risks، and supports the ethical and sustainable use of artificial intelligence.