Abstract

This paper presents an institutional case study of the Central Bank of Malta (CBM) Statistics Department’s programme to formalise data governance. The central finding is that strong technical data capabilities do not automatically translate into mature organisational data governance. A structured Data Maturity Assessment (DMA) found comparatively strong performance in systems, data management and data protection, alongside weaker results in knowing the data held, setting data direction and taking responsibility for data. These gaps provide the analytical basis for a four-pillar response: a centralised data inventory and catalogue, maturity assessment, data strategy and governance framework. Artificial intelligence (AI) was used in a deliberately narrow supporting role to generate candidate table descriptions from technical metadata using a locally deployed large language model. In response to the limitations of such metadata-only generation, the revised approach treats the model as a replaceable component, introduces domain grounding and an abstention option for ambiguous inputs, and positions expert validation as a necessary control rather than assuming generated descriptions are authoritative. The paper also proposes measurable implementation indicators that link governance interventions directly to diagnosed maturity gaps. The case suggests that sustainable data governance in statistical organisations requires the combined development of visibility, accountability, strategic direction and proportionate AI controls, while institution-wide scaling depends on cross-functional arrangements and senior organisational sponsorship.