Developing a Web-based Tool for Detecting Deceptive Designs in Cookie Banners
Braullo Jose A. Jo, Shanea J. Olino, Ligaya Leah Figueroa, Ma. Rowena Solamo, Rommel P. Feria · Atlantis Highlights in Computer Sciences/Atlantis highlights in computer sciences · 2025
Deceptive designs, also known as dark patterns, are user interface tricks websites and applications use to manipulate user behavior and collect data without informed consent.These patterns include misleading language, asymmetrical options, and hidden information.Websites commonly manifest these deceptive designs in cookie banners to track user behavior.Given the sensitive nature of the data cookies may contain, it is crucial to flag and address potential manipulations of user consent through deceptive designs in cookie banners.This study builds on Ariadne, a browser extension developed by Adorna et al. that identifies deceptive patterns in cookie banners.To improve on their work, this study presents VeraCookie, a web application that has similar functionality but is compatible with all browsers and devices.This application integrates advanced machine learning models, including a Random Forest Classifier for assessing language clarity and a Vision Transformer (ViT) model for evaluating the symmetry of the options present in a cookie banner.Results show that VeraCookie outperforms the previous tool, achieving higher accuracy and better user experience.This study aims to demonstrate the effectiveness of VeraCookie in defense against deceptive designs in cookie banners.