Machine Learning for Web Vulnerability Detection: The Case of Cross-Site Request Forgery

Abstract

Cross-site request forgery (CSRF) vulnerabilities pose a significant threat to web application security, enabling attackers to execute unauthorized actions on behalf of authenticated users. Conventional CSRF detection methods, such as manual code review and static analysis, are often time-consuming, error-prone, and inefficient. Proposes Mitch, a novel machine learning (ML)-based solution for the black-box detection of CSRF vulnerabilities. Mitch employs supervised learning, trained on a comprehensive dataset of HTTP requests and responses, to effectively identify security-sensitive HTTP requests and uncover CSRF vulnerabilities within them. Rigorous evaluations on a diverse set of real-world web applications demonstrate Mitch's remarkable ability to detect CSRF vulnerabilities with high accuracy, outperforming traditional methods. Mitch's automated nature eliminates the need for manual code review and static analysis, saving time and effort while reducing the risk of human error. Additionally, Mitch's scalability allows seamless integration into continuous integration and continuous delivery (CI/CD) pipelines, enabling continuous security monitoring and vulnerability detection. Mitch's efficacy extends beyond detecting known CSRF vulnerabilities. Its ability to identify patterns and relationships enables it to uncover obscure CSRF vulnerabilities that may have been overlooked by traditional methods, including zero-day vulnerabilities. In conclusion, Mitch emerges as a powerful tool for enhancing web application security, offering a comprehensive and automated solution for detecting CSRF vulnerabilities. Its ability to handle complex web applications, uncover hidden CSRF vulnerabilities, and integrate into CI/CD pipelines makes it an indispensable tool for web security professionals. Mitch's adoption has the potential to significantly reduce the risk of CSRF attacks and safeguard sensitive user data. We propose a methodology to leverage machine learning (ML) for the detection of web application vulnerabilities. We use it in the design of Mitch, the first ML solution for the black-box detection of cross-site request forgery vulnerabilities. Finally, we show the effectiveness of Mitch on real software. In this project, we propose a methodology to leverage Machine Learning (ML) for the detection of web application vulnerabilities. Web applications are particularly challenging to analyses, due to their diversity and the widespread adoption of custom programming practices. ML is thus very helpful for web application security it can take advantage of manually labeled data to bring the human understanding of the web application semantics into automated analysis toolsMitch allowed us to identify 35 new CSRFs on 20 major websites and 3 new CSRFs on production software. Keywords: Mitch, CSRF, CI/CD pipelines, Security Token Service (STS), Same-Origin Policy (SOP).