General terms and conditions are of significant importance in today’s economy. They define to a large extent the circumstances under which transactions are conducted and services are offered. In light of the large number of terms of service agreements entered into today, as well as their complexity, it does not seem feasible for individual consumers to monitor their implications in detail. In fact, several studies indicate that few consumers actually read the many contracts they conclude online. It is, thus, of vital interest how Artificial Intelligence and software applications may be used to effectively communicate the content of terms of services or highlight their potentially controversial elements in order to protect consumers. Hence, the goal of this thesis is to investigate possibilities of using Machine Learning to automatically classify individual terms of such agreements according to their topic. A large part of the project is concerned with devising a suitable categorization as well as building and labeling an adequate corpus which Supervised Learning algorithms can be trained with. The theoretical bases of various Machine Learning algorithms are elaborated and their ability to accurately identify a clause’s topic is experimentally investigated. The results obtained during the project and the algorithms’ observed behavior as well as the potential causes are presented and discussed.