The University AI Sovereignty Index: A Multidimensional Method for Assessing Institutional Digital Autonomy

Abstract Universities are adopting artificial intelligence through cloud platforms, enterprise assistants, learning systems, research tools, and administrative analytics, yet most institutions lack a method for judging whether these arrangements preserve meaningful digital autonomy. This methodological paper develops the University AI Sovereignty Index (UASI), a multidimensional assessment of an institution's practical capacity to choose, govern, scrutinise, and, when necessary, exit AI dependencies. The index was constructed through an integrative synthesis of digital sovereignty, platform governance, responsible AI, learning analytics ethics, composite-indicator methodology, and higher-education policy research. UASI contains eight dimensions and 40 evidence-rated indicators: strategic authority, infrastructural control, data stewardship, model assurance, pedagogical autonomy, human capability, ecosystem negotiability, and collective rights. Scores use a weighted geometric aggregation that limits compensation between dimensions, supplemented by critical safeguards for high-consequence systems. The OCAET test - orient knowledge, activate cognition, preserve agency, support evaluation, and enable transfer - anchors the pedagogical dimension. Worked hypothetical cases show why an institution with strong infrastructure but weak contestability should not receive the same judgment as a balanced institution with a similar arithmetic mean. The paper provides a scoring rubric, evidence protocol, sensitivity procedure, reporting dashboard, and staged validation programme. UASI is presented as a falsifiable methodological proposition rather than a validated ranking. Its purpose is diagnostic: to make hidden dependencies discussable, reveal capability gaps, and support better institutional choices without confusing sovereignty with technological isolation.