Knowledge graphs have become increasingly popular supplemental information because they represented structural relations between entities. Knowledge graph embedding methods (KGE) are used for various downstream tasks, e.g., knowledge graph completion, including triple classification, link prediction. However, the knowledge graph also includes much sensitive information in the training set, which is very vulnerable to privacy attacks. In this paper, we conduct such one attack, i.e., membership inference attack, on four standard KGE methods to explore the privacy vulnerabilities of knowledge graphs. Our experimental results on four benchmark knowledge graph datasets show that our privacy attacks can reveal the membership information leakage of KGE methods.