Artificial Intelligence-Driven Fraud Detection, Deepfake Prevention, and Cybersecurity Resilience in the U.S. Banking Sector: An Integrated Review of Emerging Capabilities, Explainable AI, and Real-Time Threat

The rapid digital transformation of the banking industry has increased its exposure to sophisticated financial fraud, identity theft, cyberattacks, and emerging threats associated with generative artificial intelligence (AI). Deepfake technologies, synthetic identities, voice cloning, and manipulated audiovisual content have created additional challenges for traditional banking authentication and fraud prevention systems. Artificial intelligence offers promising opportunities to strengthen financial cybersecurity through intelligent fraud detection, multimodal deepfake identification, real-time threat monitoring, and explainable decision-making. This review synthesizes two empirical studies examining AI-driven deepfake detection and fraud prevention in the U.S. banking sector. The first study investigates the relationships between multimodal deepfake detection, real-time detection, explainable AI, deepfake threat detection effectiveness, and digital banking cybersecurity resilience. The second examines AI-driven fraud detection capability and the mediating roles of Explainable AI Trust and Real-Time Fraud Response Capability in strengthening banking cybersecurity resilience. Both studies employ quantitative research designs and Partial Least Squares Structural Equation Modeling (PLS-SEM) to evaluate organizational-level relationships. Their findings demonstrate that AI-driven detection capabilities can contribute significantly to cybersecurity resilience. However, explainability, organizational trust, and the effective integration of AI-generated intelligence into institutional security processes remain essential. This review identifies important research opportunities involving multimodal AI, trustworthy fraud detection, human-AI collaboration, privacy-preserving technologies, and adaptive cybersecurity frameworks. It concludes that effective AI-powered banking security requires an integrated organizational approach that combines intelligent detection, transparent decision-making, operational preparedness, and continuous threat adaptation.