CAI Cuihong
Deputy Director of Center for Global AI Innovative Governance; Professor of the Institute of International Studies, Fudan University
The central challenge of global AI governance lies not in a lack of institutions, but in a structural mismatch between institutional expansion and governance capacity. In recent years, AI governance frameworks have proliferated across all levels, yet the number of incidents has risen concurrently, indicating that institutional growth has not automatically translated into governance effectiveness. This mismatch manifests in two aspects. On the one hand, mechanism expansion coexists with a capacity vacuum — many countries, especially in the Global South, lack the practical capacity to implement governance frameworks. On the other hand, there is a power misalignment: AI governance is state-driven, but AI operations are corporation-centric, with computing power, data, and safety standard-setting concentrated in major technology companies. Therefore, the critical question in AI governance is not how many mechanisms have been established, but who truly possesses governance capacity and how that capacity is distributed. Specifically, governance capacity should be understood as comprising three layers: technical capability, institutional capability, and credibility.
