Systems, computer products, and methods are described herein for improvements for identifying if an interaction between a user and an entity have been misappropriated. Instead of analyzing each interaction as the interactions are entered into between various users and the associated entities, a specific federated user model may be created for each user in order to predict the likelihood that interactions of the user may be misappropriated. In order to create the federated user model, a quantum optimizer may assign qubits to various inputs and analyze the inputs to create a federated user model that best predicts the user's interactions that may be misappropriated. As the inputs change, the quantum optimizer may be used to morph the federated user model in order to improve upon the ability of the specific federated user model in determining the likelihood that the user's interactions are misappropriated.