Detection and Prevention of Data Leakage in Transit Using LSTM Recurrent Neural Network with Encryption Algorithm

Abstract

As the amount of data generated by organizations continues to multiply, there is a need to ensure the security and transmission of these data at any stage that they may be at. Data breaches have constituted a large loss when it comes to the integrity of an organization and their sensitive information being leaked or authorized by an unauthorised user. This research implements a solution to detect and prevent data leakage in transit using LSTM and the AES-256 encryption algorithm. Using a real-world news dataset, the study introduced a solution using LSTM to train the dataset and AES-256 encryption to encrypt the data based on the trained dataset. Test data was then passed to confirm the accuracy of the model and detect if there was a match on the news dataset. From the results of the study, the model was able to detect the sample or test data accurately as leaked data with an accuracy of 93.7%. The importance of developing a system to mitigate these risks is to help provide organizations with confidence in how their confidential information is transmitted.