Data Leakage Detection Using Cloud Computing

Abstract

Data leakage refers to the unintentional or accidental disclosure of confidential or sensitive information to unauthorized individuals or entities. This poses a significant challenge for businesses, as the frequency and costs associated with data leakage incidents continue to rise. The issue of data leakage is compounded by the lack of regulation and monitoring for transmitted data, including emails, instant messaging, file transfers, and web forms. This makes it difficult to track and control the flow of data to its intended recipients. In our research, we focus on a specific problem: a data distributor entrusts sensitive data to a group of agents who are deemed trustworthy. However, some of the data is leaked and found in unauthorized locations, such as the internet or unauthorized devices. The distributor must assess the probability that the leaked data originated from one or more of the entrusted agents, rather than being obtained independently through other means. To address this challenge, we propose data allocation strategies among the agents that aim to improve the likelihood of identifying data leakages. Importantly, our methods do not rely on modifying the released data through techniques like watermarks. In certain cases, we also explore the possibility of introducing realistic but fabricated data records to further enhance our ability to detect leaks and identify the responsible party.