Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{209648,
        author = {Rahul Jadhav and Mahesh Rane},
        title = {AI Cyber-Shield: An Explainable Machine-Learning Framework for Phishing and Quishing Detection},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {5},
        pages = {2881-2884},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=209648},
        abstract = {Phishing remains a major cybersecurity threat in which attackers use fraudulent websites, emails, and messages to steal passwords, banking information, and other sensitive data. A newer variant, Quishing or QR phishing, hides malicious destinations inside QR codes and makes manual inspection difficult for ordinary users. This paper presents AI Cyber-Shield, a web-based phishing and Quishing detection framework that unifies URL analysis, QR-code decoding, and email analysis in a single system. The proposed system combines rule-based security indicators with a Python-based machine-learning layer supporting Random Forest, XGBoost, Support Vector Machine (SVM), and ensemble classification. URL features include HTTPS usage, domain structure, URL length, suspicious keywords, excessive symbols, IP-based URLs, subdomains, shorteners, redirects, brand-related spelling errors, and possible typo squatting. QR images are decoded without opening or executing their embedded links, while email analysis examines sender information, subject, message body, embedded links, urgent language, account-verification requests, impersonation indicators, and other social-engineering patterns. The framework produces a SAFE, SUSPICIOUS, or PHISHING classification, a risk score from 0 to 100, detected warning indicators, an explanation, and a recommended action. Machine-learning performance is intended to be evaluated using Accuracy, Precision, Recall, F1-Score, and a Confusion Matrix. The design emphasizes unified detection and understandable results rather than relying on a single input type or an unexplained binary prediction.},
        keywords = {Phishing Detection, Quishing, QR Phishing, Cybersecurity, Machine Learning, Random Forest, XGBoost, Support Vector Machine, Ensemble Classification, Typosquatting, Explainable Security.},
        month = {October},
        }