The web is rich of third-party services which use HTML5 and JavaScript snippets, frequently obfuscated, in order to uniquely identify the user and monitor him when surfing the web. This approach, called fingerprinting, constitutes a thread for users' privacy and companies' security. In the thesis it has been conducted a census of the techniques used by tracking services to fingerprint users, the APIs which constitute a source for the unique identification have been identified and it has been developed an automatic, machine learning based detection system for scripts which perform fingerprinting.