AI Governance Mechanisms for a Sustainable Digital Transformation: A Case Study of a Swedish Public Organization

Sammanfattning

Introduction: Artificial Intelligence (AI) has emerged as a key driver of sustainable digital transformation in modern organizations. However, the effective use of AI requires well-defined AI governance mechanisms to ensure that AI systems are implemented responsibly and aligned with organizational objectives. In the absence of effective AI governance, poorly managed AI systems can undermine transformation efforts and create significant legal, ethical, and reputational risks for organizations. Despite the growing importance of AI governance, existing research provides limited insight into what AI governance mechanisms are implemented in practice, particularly within public-sector contexts. To address this gap, this study investigates the AI governance mechanisms implemented at the organizational level to support sustainable digital transformation. Research Question: The primary research question of this study is: What AI governance mechanisms are implemented at the organizational level to support sustainable digital transformation in a Swedish public organization? Method: The qualitative single case study strategy has been used in this study, and data were collected by six semi-structured interviews with the employees in strategic, managerial, and technical positions in the selected Swedish public sector organization, and were complemented by internal documents related to AI policy and sustainable AI. The collected data was analyzed using thematic analysis to identify key AI governance mechanisms. Results During thematic analysis, eight key AI governance mechanisms were identified within the case organization. These key mechanisms are grouped into eight themes: Structural and Organizational AI Governance Mechanisms, Risk Management and Regulatory Compliance Mechanisms, Ethical and Responsible AI Governance, Data and Model Lifecycle AI Governance, Organizational Oversight, Audit, and Continuous Improvement, Human Oversight and Accountability Mechanisms, Organizational Capability and AI Literacy, and External Collaboration and Stakeholder-Oriented AI Governance. Across these themes, 20 sub-mechanisms were identified. The majority of these AI governance sub-mechanisms correspond to those discussed in existing literature. However, two sub-mechanisms are identified as context-specific contributions of this study. For the key theme of Structural and Organizational AI Governance Mechanisms, Integration into existing organizational governance processes is identified as a new sub-mechanism. Additionally, for the key theme of Organizational Capability and AI Literacy, External Knowledge Networks and Capability Building is identified as the second new sub-mechanism. The remaining sub-mechanisms were found to be consistent with prior research. Discussion: The identified AI governance mechanisms highlight how AI governance is implemented in practice within a Swedish public-sector organization. Since most mechanisms align with existing literature, the findings suggest that established AI governance approaches are applicable across organizational contexts. However, the study shows that their effectiveness depends on how they are integrated into existing processes and supported by organizational structures. The eight key mechanisms of AI governance provide evidence that practical AI governance is inherently multi-dimensional in nature, requiring the coordination of structural, regulatory, ethical, technological, and capacity-building aspects simultaneously, not individually. The study also finds that there are two new additional sub-mechanisms, which expand upon existing governance approaches in prior studies. These new submechanisms suggest that public sector organizations do not just follow conventional practices but incorporate AI governance in their organizational processes and enhance their capabilities through knowledge networks outside the organization. Based on the eight key AI governance themes identified from the findings, this study proposes an AI governance framework for sustainable digital transformation, organised across three layers: enabling, operational, and accountability, and supported by cross-cutting enablers. The proposed framework contributes by demonstrating that sustainable digital transformation emerges from the integrated functioning of AI governance layers rather than from individual mechanisms alone. This suggested framework can serve as a useful resource for both researchers and practitioners regarding the development and deployment of AI governance mechanisms in public sector organizations. Although this study is based on a single public-sector case, the findings can be generalized to similar public-sector organizations operating in comparable regulatory and institutional contexts. Future research should explore AI governance mechanisms across multiple public organizational contexts.