Publication Details
Abstract
The rapid uptake of artificial intelligence (AI) in credit scoring, investment decisions, and risk management has outpaced the ability of regulators and bank boards to validate it, a concern that is acute for systemically important banks (SIBs), where an inadequately governed model can transmit losses across the financial system. Existing diagnostic tools are either technology-focused AI maturity models or principle-based supervisory frameworks such as the NIST AI Risk Management Framework, ISO/IEC 42001, and the Basel Committee's D-SIB principles, none of which is designed to assess whether a specific SIB has the organisational capability to understand, control, and, if necessary, halt its own AI-driven credit and investment processes before they cause financial damage. This paper introduces the Composite AI Governance Maturity Index (CAGMI), built on eight weighted governance blocks, thirty-two threshold-scored indicators, and non-compensable “red-line” restrictions, and applies it as a documentary and illustrative exercise to Uzbekistan’s seven SIBs, which held 62.4% of banking-sector assets and 65.9% of the credit portfolio as of 1 January 2026. A structured review of public disclosure finds uneven institutionalisation, with strategy and personnel commitments far outweighing independent validation and financial-impact monitoring. An explicitly simulated illustrative panel exercise demonstrates how CAGMI could be linked econometrically to bank-level risk and performance outcomes once internal data become available, and the paper closes with policy recommendations for the Central Bank of Uzbekistan and for other digitally and AI-transforming emerging-market banking systems.