A method for identifying hidden information leak channels due to incorrect updates of database management system kernel programs

This article examines the problem of vulnerabilities that arise during software updates in isolated corporate systems running Astra Linux. A possible method for implementing an attack to gain root access through a modified software update package is demonstrated. A multi-layered method for detecting update package modifications is proposed, including static, behavioral, and signature-based analysis using machine learning methods and algorithms. This method significantly improves the security of the software update process in network isolation.