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{175705,
        author = {Ch Jaswanth Kumar and K Venkata Ratnam and D Asha Jyothi and B Aravindasai},
        title = {Phishing Website Detection Using Machine Learning via Browser Extension and Mobile Application},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {11},
        pages = {4323-4326},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=175705},
        abstract = {This paper introduces a novel, multi-layered approach to phishing detection, deployed across web browsers and Android platforms, offering real-time protection against advanced phishing threats. The system utilizes a hybrid model that combines Random Forest (RF) classification as the first layer of defense with Large Language Models (LLMs) as a second, more sophisticated layer to enhance detection accuracy. The Random Forest classifier analyzes key URL features such as domain structure, URL length, presence of suspicious keywords, and HTTPS status to provide an initial prediction with high speed and efficiency. Once a potential threat is detected or flagged as ambiguous, the LLM performs a deeper semantic analysis of the webpage content and associated metadata, identifying subtle linguistic patterns, inconsistencies, and suspicious behaviors that may escape traditional detection models. This layered detection mechanism ensures high accuracy, capturing zero-hour phishing attempts and reducing false positives significantly. The system is implemented through browser extension sand Android accessibility services, allowing seamless cross-platform functionality to protect users across different environments. In real-world deployment scenarios, this hybrid model demonstrated exceptional performance, achieving an impressive 97.8% accuracy and maintaining a false-positive rate of just 0.3%. By combining the strengths of machine learning and natural language understanding, our system effectively mitigates the risks of phishing attacks, offering a comprehensive and adaptive security solution across web and mobile platforms.},
        keywords = {phishing detection, machine learning, browser extension, mobile application, cybersecurity, random forest, large language models, zero-hour detection, cross-platform protection, hybrid model.},
        month = {April},
        }