A Gaussian Mixture Model is used for clustering, assuming each cluster follows a Gaussian distribution. The data actually contains a uniform rectangular cluster and a Gaussian cluster. What is the likely failure?