The frequency of cybersecurity events and data breaches is escalating, resulting in substantial financial liabilities for businesses. Contemporary organizations rely heavily on network connections to facilitate efficient operations in diverse locations. Covert timing channels (CTCs), a sophisticated cyber threat, involve the surreptitious embedding of sensitive information into regular network traffic, thus establishing concealed communication pathways that are difficult to detect. This study proposes a novel approach for detecting CTCs using machine learning (ML), which significantly advances the state-of-the-art in data breach detection and information leakage through CTC. Our approach introduces a unique integration of Unicode transformation and ML that enhances the detection capability by analyzing inter-packet timing data. In particular, our findings indicate that the Gradient Boost Classification model yields an accuracy rate ranging from 73% to 78%, while the Random Forest model (RF) also exhibits significant accuracy.