As the threat of data leakage by internal employee increases, the enterprises are applying security solutions such as Data Loss Prevention (DLP) and Digital Rights Management (DRM) to prevent data leakage. However, since data leakage method also become more diverse, it makes more difficult to preventing data leakage. In this paper, we proposed the method to detecting the data leakage by monitoring working activities of inter employee through the security log. The proposed method draw graph using the security log collected from the behavior of internal employee using data leakage scenario and detects whether the data is leaked through the deep learning algorithm which shows excellent performance in processing large data. Proposed method shows 95% accuracy regardless of leaked data size. Also, proposed method shows 97% accuracy regardless of the number of attempts to obtain data.