Where universities place generative AI governance: A comparative document analysis of institutions in Korea, Japan, and China

Abstract

This study examines where universities place generative-AI governance and whether cross-national differences partly reflect differences in institutional document type rather than only substantive policy content. A sampling frame of 36 universities in Korea, Japan, and China was fixed before document retrieval. Official and verifiable instruments were coded for document genre, issuer level, governance mode, timing, and citation anchors. Thirteen records received full close reading, and structural variables were independently coded with adjudication. Thirty-one institutions had verified instruments, two were partially verified, and three had no confirmed public instrument. Korean and Japanese source-bearing primary cases were concentrated in teaching and learning, whereas the Chinese corpus additionally contained thesis or degree instruments and one AI system-security instrument. This structural contrast persisted after excluding China’s specialised stratum, although its magnitude decreased. Cross-national policy comparison should establish document genre, organisational level, and source provenance before comparing provisions. Universities can improve governance coherence by matching AI rules to the responsible institutional unit, recording issue and revision histories, and distinguishing teaching guidance, degree control, research integrity, and system-security functions.