Enhancing Cloud Security with Machine Learning: Tackling Data Breaches and Insider Threats
Sreejith Sreekandan Nair, Govindarajan Lakshmikanthan, J. S. Ray Parthasarathy, Savithramma P. Dinesh‐Kumar, Kavitha Shanmugakani, B. Jegajothi · 2025
Cloud computing remains susceptible to internal threats and data compromises, notwithstanding its foundational role in contemporary business operations within this age of automation. To mitigate this concern, a machine learning framework is suggested for the preemption of prospective threats. Most, breaches of organizational internal systems are concerns about user non-compliant behaviors aimed at messing up activity logs, which can be combatted using supervised models and change detection. There seems to be a class of threats that go undetected that the authors would refer to as behaviour patterns that do not follow standard operating behaviour for business. An interface of RF and LSTMs would be able to establish common patterns and knit rule-based threats to create business insights and acceptable threats. User and System exhaustion of RF was used to minimise covert system identifiers of users to examine access pattern data streams, which were unlimited in dimensional centers. Evaluation of detection of anomalous content, such as R- scoring, classification of log events as generic clustering, etc. It could prevail on the challenges of log records analysis and the checkpoints of Structural Semantic Communication principles. It can free up a range of security checkpoints by glaring at user access limits and ensuring multi-channel compliance. This approach would significantly strengthen data security on the Cloud by targeting high-risk actions across oriented sections of breaching data storage centers.