Home / Current Issue / Paper 1712408
AI Shield: A Hybrid Machine Learning and Deep Learning Approach for Detecting Malicious URLs
Subject area: Science,Engineering and Technology · Area of research: Machine Learning
DOI: https://doi.org/10.64388/IREV9I5-1712408
Abstract
In the modern digital era, the exponential growth of online activities has resulted in an alarming increase in cyber threats, especially phishing and malicious URLs. These threats exploit user trust and vulnerabilities in online systems to steal sensitive credentials and financial information. Traditional defense mechanisms such as blacklist-based filters and static rule-based systems are insufficient, as attackers continuously evolve their techniques to bypass detection. To overcome these limitations, this paper presents AI Shield, a hybrid detection framework that integrates Machine Learning (Decision Tree) and Deep Learning (LSTM) models for the intelligent classification of URLs as safe or malicious. The Decision Tree model performs rule-based lexical analysis, while the LSTM captures sequential dependencies in URL structures, enabling deeper behavioral understanding. Experimental evaluations on the PhishTank 2024 dataset demonstrate that the proposed hybrid approach achieves a detection accuracy of 95%, surpassing standalone ML and DL models. The hybridization approach enhances adaptability, scalability, and real-time detection capability, making AI Shield a robust solution for phishing mitigation in modern cybersecurity infrastructures.
Keywords
Phishing Detection, Malicious URL, Hybrid Model, Decision Tree, LSTM, Cybersecurity, Deep Learning, Machine Learning.
How to cite this paper
@article{1712408,
author = {Chaithra SG, Bhagyashree, Irfan Khan, Chaithanya Lokesh, Bharath. GV},
title = {AI Shield: A Hybrid Machine Learning and Deep Learning Approach for Detecting Malicious URLs},
journal = {Iconic Research And Engineering Journals},
year = {2025},
volume = {9},
number = {5},
pages = {2104-2110},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1712408.pdf},
abstract = {In the modern digital era, the exponential growth of online activities has resulted in an alarming increase in cyber threats, especially phishing and malicious URLs. These threats exploit user trust and vulnerabilities in online systems to steal sensitive credentials and financial information. Traditional defense mechanisms such as blacklist-based filters and static rule-based systems are insufficient, as attackers continuously evolve their techniques to bypass detection. To overcome these limitations, this paper presents AI Shield, a hybrid detection framework that integrates Machine Learning (Decision Tree) and Deep Learning (LSTM) models for the intelligent classification of URLs as safe or malicious. The Decision Tree model performs rule-based lexical analysis, while the LSTM captures sequential dependencies in URL structures, enabling deeper behavioral understanding. Experimental evaluations on the PhishTank 2024 dataset demonstrate that the proposed hybrid approach achieves a detection accuracy of 95%, surpassing standalone ML and DL models. The hybridization approach enhances adaptability, scalability, and real-time detection capability, making AI Shield a robust solution for phishing mitigation in modern cybersecurity infrastructures.},
keywords = {Phishing Detection, Malicious URL, Hybrid Model, Decision Tree, LSTM, Cybersecurity, Deep Learning, Machine Learning.},
month = {November},
doi = {https://doi.org/10.64388/IREV9I5-1712408}
}