Abstract

As interest in generative AI has rapidly increased in recent years, there is a growing willingness to adopt large-scale AI in the public sector. However, precise implementation methods remain elusive due to various issues related to the introduction of large-scale AI. This study proposes a public data sovereign cloud to address two main challenges when introducing large-scale AI in the public sector: 1) compliance with information disclosure laws and regulations and 2) operation of closed networks. To this end, we examine the concept of a sovereign cloud and propose a definition and implementation strategy for a public data sovereign cloud. Specifically, we present application strategies focusing on construction methods using public-private partnership (PPP) models, network segregation methods according to data grades, and enhancement of the role of Managed Service Providers (MSPs). The results of this study are expected to contribute to policy formulation for introducing and utilizing large-scale AI in the public sector.