An AI-Powered Approach to Real-Time Phishing Detection for Cybersecurity
Year: 2025
Type: article
Abstract: Email phishing is one of the most significant cybersecurity threats in the digital era, leading to financial losses and data breaches. This study presents an AI-based real-time phishing detection system that employs machine learning techniques such as Logistic Regression, Random Forest, Support Vector Machine (SVM), CatBoost, and XGBoost. The dataset, sourced from Kaggle, includes phishing and safe emails to train and evaluate these models. Preprocessing methods including text normalisation, stemming, and feature extraction were used to improve detection accuracy. The findings show that XGBoost and CatBoost had the highest accuracy of 98%, outperforming other models. Logistic Regression and Random Forest followed closely with 97% accuracy, while SVM had 96%. The findings emphasise AI's effectiveness in detecting phishing emails and the importance of continuous cybersecurity measures. This study enhances email security by demonstrating the effectiveness of AI-driven phishing prevention mechanisms
Language: English
Topic: Spam and Phishing Detection
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4408811456