V3I3P54

AI-DRIVEN FINANCIAL RISK DETECTION IN ENTERPRISE INFORMATION SYSTEMS: Implications, Architecture, and Policy for the United States Financial Ecosystem

Ayomipo Alademehin^1*, Bernice Blay-Miezah²

Abstract

Artificial intelligence is reshaping how financial institutions detect, measure, and respond to risk. This paper presents a comprehensive, critically grounded analysis of AI-driven financial risk detection within enterprise information systems (EIS), with a specific focus on implications for the United States financial ecosystem. Drawing on a systematic review of peer-reviewed literature, regulatory guidance, and documented institutional deployments, the study examines eight categories of financial risk and maps corresponding AI techniques to each domain, evaluating empirical performance evidence for every mapping.

Results drawn from the published literature indicate that AI-integrated EIS platforms substantially outperform rule-based approaches across all major risk categories. Ensemble architectures report AUC ROC scores approaching 0.98 in fraud detection benchmarks, while federated learning models achieve credit risk accuracy within 2.3 percentage points of centralised alternatives while satisfying Gramm Leach Bliley Act data sharing constraints (Zheng et al., 2022). These figures are drawn from previously published studies and should be interpreted as indicative of directional improvement rather than as universally generalisable results. Performance varies across datasets, institutional contexts, and regulatory environments.

The paper introduces two original contributions: a five-level EIS AI Risk Detection Maturity Model (EARM) that enables institutions to assess and develop their AI integration capabilities, and a phased 24-month implementation roadmap calibrated to U.S. regulatory requirements. Critical challenges, including model explainability, algorithmic fairness, adversarial robustness, and regulatory fragmentation, are systematically examined with evidence-based mitigations. The paper concludes with policy recommendations directed at the Federal Reserve, Office of the Comptroller of the Currency, FinCEN, and U.S. Congress, and identifies priority directions for future empirical research.

Keywords:

Artificial Intelligence, Financial Risk Detection, Enterprise Information Systems, Machine Learning, Federated Learning, Credit Risk, Fraud Detection, Systemic Risk, U.S. Financial Regulation, Explainability, Algorithmic Fairness