Analysis of Factors Shaping AI Sovereignty Perception: The Differential Effects of Publicness and Privacy

This study aimed to clarify individuals’ perceptions of AI sovereignty within the tension between AI publicness and privacy protection. For this study, a structural equation modeling analysis (using the StataNow/SE 19.5 program) was conducted on 210 adults. The results showed that only AI privacy significantly positively influenced perceptions of AI sovereignty. This suggests that privacy protection must precede the realization of AI sovereignty. Multi-group and moderating analysis revealed that women tended to perceive AI sovereignty through the shared responsibility structure of the public sector. Furthermore, the interaction between “expectations for explanations” and “threat concerns” strengthened perceptions of sovereignty, suggesting that specific explanations and experiences can offset perceived threats regarding AI. This study suggests that future AI policies should be designed around the introduction of privacy-preserving technologies (PETs), expanded explainable AI (XAI) education, and the provision of secure services led by the public sector.