Systemic Opacity Risk: Macroprudential Dimension of AI Governance in Banking [V.02]

Abstract— The integration of artificial intelligence into banking has intensified reliance on decision systems whose internal logic may not be fully reconstructible. Existing analyses have addressed opacity primarily at the institutional level, focusing on governance, explainability, and model oversight. This paper introduces Systemic Opacity Risk (SOR) as a macroprudential dimension of AI-driven finance. SOR is defined as the risk that the aggregation and correlation of institutional opacity—even where individually contained within tolerance thresholds—may impair the financial system’s collective capacity to reconstruct decisions under systemic stress. The framework remains conceptual. It does not propose quantification or immediate regulatory intervention. Its contribution lies in identifying a dimension of systemic vulnerability linked not to the propagation of financial losses, but to the preservation of system-level intelligibility under stress. [This expanded version develops the structural and macroprudential dimensions of Systemic Opacity Risk introduced in the earlier conceptual paper circulated in March 2026; it builds and extends the author’s prior research on AI explainability, prudential architecture, and tolerance for opacity]