Real-Time Cyber Risk Assessment in US Cloud-Native Open Banking APIs: A Machine Learning-Driven Threat Analytics Framework for Fraud Prevention and Infrastructure Protection

• Adaan Ahsun

  Covenant University

  Author

The rapid migration of US financial institutions toward cloud-native architectures and open banking ecosystems has fundamentally expanded the cyber attack surface, with APIs now serving as both the engine of innovation and the primary vector for exploitation. Despite the proliferation of machine learning-based fraud detection systems, existing approaches remain constrained by static rule engines, inadequate dynamic threat responsiveness, and limited integration of real-time risk intelligence across heterogeneous data sources. This study addresses these gaps by proposing and validating a novel Machine Learning-Driven Threat Analytics Framework (ML-TAF) that integrates multimodal data fusion, a hybrid ResNet-LSTM architecture for spatial-temporal feature extraction, dynamic security assessment via Bayesian networks, and multi-criteria decision-making for actionable security strategy recommendations. Using a comprehensive dataset comprising 1.2 million transaction records, 300,000 user sessions, and network telemetry from US financial institutions, the framework achieved 89.4% accuracy in real-time fraud detection, a 73.1% risk reduction rate, and demonstrated 40% lower inference latency compared to baseline models. The framework provides a replicable, privacy-preserving, and interpretable solution for financial institutions seeking to operationalize proactive cyber risk management in API-driven ecosystems.

• pdf

Copyright (c) 2026 Adaan Ahsun (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.