ECONOMIC EFFICIENCY AND SECURITY ISSUES IN THE IMPLEMENTATION OF ARTIFICIAL INTELLIGENCE TECHNOLOGIES IN COMMERCIAL BANKS: PROBLEMS AND SOLUTIONS

Abstract

The accelerated adoption of artificial intelligence (AI) technologies in commercial banking has reshaped the operational logic, risk architecture, and competitive boundaries of financial institutions worldwide. This article examines the dual nature of AI integration—its economic efficiency dividends and its emerging security vulnerabilities—through a synthesis of recent empirical evidence from developed and developing markets, with particular attention to the case of the Republic of Uzbekistan. Drawing on data from international financial supervisors, peer-reviewed studies, and recent industry reports, the paper finds that AI deployment generates measurable gains in operational productivity, credit risk assessment accuracy, customer-service automation, and cross-channel personalization, while simultaneously introducing previously unknown risks linked to model opacity, adversarial attacks, deepfake-enabled fraud, third-party dependencies, and data-governance failures. The analysis proposes a four-tier framework—Data integrity, Algorithmic accountability, Operational resilience, and Supervisory convergence (DAOS)—as a structured response to the trade-offs banks face. Empirical illustrations from Uzbek commercial banks demonstrate that local AI deployment improves credit-scoring accuracy by 14–16 percentage points relative to traditional models, but remains constrained by talent shortages, fragmented data ecosystems, and incomplete regulatory infrastructure. The paper concludes that economic efficiency and security are not competing objectives but complementary outcomes that depend on the institutional maturity of governance frameworks, the integration of cybersecurity into AI lifecycle management, and the alignment of national supervisory practice with emerging international standards.