The development of new internet services brings significant benefits to society. Many of these services rely on sophisticated algorithms or machine learning models, enabling them to tackle challenging tasks efficiently. Consequently, they simplify our lives by automating complex processes and spare our time by providing effective solutions.

Yet, sharing valuable machine learning or complex algorithmic solutions on the Internet exposes owners to potential risks. While not fully disclosed, the underlying program logic is utilized in a service accessible to clients for sending requests and receiving feedback. Such interactions may reveal crucial details about the solution and, if manipulated by a malicious entity, could be exploited for service theft, leading to a violation of intellectual property rights. For instance, if a service with a machine learning model underneath is under attack, a malicious client can use the service feedback to train their own machine learning model, replicating the behaviour of the original model. Subsequently, such an illegally copied model can be offered as a cheaper alternative solution, leading to profit loss for the original service provider and forming unfair competition.

To address such violations of intellectual property of service providers, in MONITAUR, we aim to develop a monitoring toolkit for tracking the traffic of service clients to detect those behaving suspiciously. We will combine the state-of-the-art approaches to prevent, in particular, machine learning model theft and integrate these detection techniques into open-source monitoring tools. 

By reducing risks of unauthorized service copying, MONITAUR aims to enhance the safety of sharing complex solutions as services on the Internet and promote a more open and secure community.