Leveraging Artificial Intelligence to Combat Cyber Threats in Financial Institutions

Abstract

The proposed study explores how artificial intelligence can be used to improve cybersecurity in financial institutions through machine learning, deep learning, and behavioural biometrics in the detection of fraud. It compares the models, including the Random Forest, Logistic Regression, Isolation Forest, and a deep-learning classifier. The mean reveals that Random Forest is better than Logistic Regression at an accuracy of 78% compared to 51% which is statistically significant (p = 3.37e-08). Nevertheless, none of the models can adequately detect fraudulent transactions indicating that it is quite challenging to identify fraud in real-time in financial ecosystems, and that AI-based detection systems still require further enhancements.