How to Enhance User Privacy Via Transparent Cookie Statements

Cookies are widely used on the Internet, but users lack sufficient knowledge to manage them. Even though the websites should inform users about the explicit purposes of their data collecting, unfortunately, many forms of cookie notices have been proven ineffective. Some of websites use deceptive user interface designed to manipulate users' behaviors and choices. In our work, we propose a set of measurement rules for checking cookie statements. We use the GPT model to score the cookie statements of 520 top-ranking websites. The results show that the score of 349 cookie statements is less than six points (the full score is ten points), and they lack important information about the use of cookies. In addition, we use the trained machine learning model to explore potential violations on these websites. Among 1,300 popular websites, the results indicate that 72.4% of them have at least one violation.