Follow Us on Google
By Rita Okoye
Digital banking has transformed financial services, but with it has come a new wave of sophisticated fraud that is costing individuals and institutions billions each year.
Nigerian researcher and Ph.D. candidate Joy Nnenna Okolo is tackling this challenge head-on with artificial intelligence (AI) tools designed to detect fraudulent activity before it causes harm.
In her publication “AI-powered fraud detection in digital banking”, Okolo explores how machine learning models can be trained to recognise suspicious patterns in financial transactions, even when fraudsters use advanced evasion techniques
. She notes that traditional rule-based fraud detection systems are often slow to adapt to new threats, leaving banks vulnerable.
“The challenge is that cybercriminals are constantly evolving their methods,” Okolo said. “Static systems can’t keep up. We need models that learn continuously and adapt in real time.”
Her approach combines supervised and unsupervised learning techniques, enabling the detection of both known and emerging fraud patterns. This includes anomaly detection algorithms that flag transactions deviating from normal customer behaviour, as well as classification models that categorise risk levels.
Okolo’s research also integrates behavioural science with cyber threat intelligence (CTI). By factoring in human-enabled breaches — such as phishing attacks and insider threats — her models address a gap in many existing fraud systems
.
One of the critical issues she highlights is the false positive rate in fraud detection. High rates can frustrate customers and increase operational costs for banks. Her model aims to reduce false positives while maintaining high accuracy, ensuring that legitimate transactions are not wrongly flagged.
In financial institutions, the stakes are high. A single breach can undermine customer trust and trigger regulatory penalties. Okolo’s work offers a proactive shield, allowing systems to respond instantly when suspicious activity is detected.
Her experience in Nigeria’s banking sector adds real-world insight to her academic research. During her years at Access Bank Plc, she implemented data analytics strategies that improved customer service ratings and reactivated dormant accounts — proving the power of targeted, data-driven interventions
.
Okolo’s AI-driven fraud detection framework has applications beyond banking. It could also be adapted for e-commerce, digital payment platforms, and cryptocurrency exchanges, where transaction volumes are high and risk patterns are constantly shifting.
She sees industry collaboration as essential to success. “No single institution has the full picture of fraud trends. Secure data-sharing frameworks can make detection models more robust,” she said.
The next phase of her research will focus on integrating these AI models into privacy-preserving computing environments, ensuring that sensitive financial data can be analysed without being exposed. This links directly to her work in federated learning, a technology she believes will reshape financial security in the next decade.
“Fraud is not going away,” Okolo said. “But with AI, we can make it harder, riskier, and far less profitable for cybercriminals.”
