AI-Enhanced SOC Operations for Deepfake and Synthetic Fraud Detection in Banking: A Comparative Study with Traditional SIEM (2018–2026)

Abstract

This study investigated the comparative effectiveness of AI-enhanced Security Operations Center (SOC) systems and traditional SIEM-based detection mechanisms in identifying deepfake and synthetic fraud in banking environments from 2018 to 2026 using a quantitative quasi-experimental longitudinal design. The analysis was conducted on a dataset of 1,250,000 cybersecurity events, including 937,500 legitimate cases (75.0%) and 312,500 fraudulent cases (25.0%), of which 187,500 (15.0%) were synthetic identity fraud, 75,000 (6.0%) deepfake-related fraud, and 50,000 (4.0%) conventional fraud. The findings indicated that AI-enhanced SOC systems achieved a higher overall detection accuracy of 94.6% compared to 82.3% for traditional SIEM systems, reflecting a mean performance improvement of 12.3 percentage points. In synthetic identity fraud detection, AI systems reached 93.4% accuracy compared to 74.8% for SIEM, while deepfake fraud detection showed 91.1% versus 69.7%. False-positive rates were significantly reduced in AI systems (4.7%) compared to SIEM (18.9%), representing a reduction of 14.2 percentage points, while false-negative rates declined from 11.7% in SIEM to 3.2% in AI systems. Response time analysis revealed that AI-enhanced SOC systems achieved an average response time of 3.8 minutes compared to 12.6 minutes in SIEM environments, indicating an improvement of 8.8 minutes. Under high event density conditions, AI systems maintained stable accuracy with variation limited to ±2.3%, whereas SIEM performance declined by up to 9.8% and experienced alert backlog increases reaching 35.6%. Statistical testing confirmed that all differences were significant (p < 0.05), with large effect sizes (Cohen’s d ranging from 0.96 to 1.41) and strong regression relationships (β up to 0.72, R² up to 0.65). The results demonstrated that AI-enhanced SOC systems provided superior accuracy, efficiency, and scalability in detecting complex and evolving fraud patterns in modern banking systems.