Artificial Intelligence in Cybersecurity: A Comprehensive Analysis of Machine Learning Applications on Phishing URLs and Malware API Calls for Devising Cyber Defense Strategies

The persistent rise of cyber threats, particularly phishing and malware attacks, has created an urgent need for innovative and robust defense mechanisms. Traditional rule-based security systems frequently struggle to adapt to rapidly changing attack vectors, creating significant vulnerabilities across organizational infrastructures. To address these challenges, this study presents a comprehensive analysis of Artificial Intelligence (AI) techniques—especially machine learning models—and their applications in detecting phishing URLs and malware API calls. By leveraging three publicly available datasets and one newly curated dataset through web scraping, this research fills critical gaps in existing literature related to adaptability, scalability, and early-stage threat detection. A systematic literature review highlights limitation of current solutions, including high false-positive rates and limited generalizability. We propose a novel, hybrid framework combining Random Forest and a Deep Neural Network to tackle these issues. Experimental evaluations span key performance metrics (accuracy, precision, recall, F1 score) and detailed statistical analyses. Results indicate the proposed approach offers enhanced detection rates and lower error margins than standalone methods, demonstrating both theoretical and practical benefits in real-world cybersecurity applications. Ultimately, this research contributes an adaptive, efficient, and scalable solution that addresses the dual challenges of phishing and malware threats, providing a foundation for improved AI-driven cybersecurity strategies in dynamic threat environments.