As large-scale graphs become more widespread today, it exposes computational challenges to extract, process, and interpret large graph data. It is therefore natural to search for ways to summarize the original graph while maintaining its key characteristics. In this survey, we outline the most current progress of deep learning on graphs for graph summarization explicitly concentrating on Graph Neural Networks (GNNs) methods. We structure the paper into four categories, including graph recurrent networks, graph convolutional networks, graph autoencoders, and graph attention networks. We also discuss a new booming line of research which is elaborating on using graph reinforcement learning for evaluating and improving the quality of graph summaries. Finally, we conclude this survey and discuss a number of open research challenges that would motivate further study in this area.