Efficient scheduling of security-sensitive tasks in cloud-edge computing systems presents a complex challenge due to the need to balance resource constraints, security requirements, and the dynamic nature of both edge and cloud environments. Existing task scheduling approaches often struggle with performance optimization, leading to inefficient resource utilization, slow task execution, and heightened security risks during offloading. These limitations stem from the inability to effectively handle resource heterogeneity, workload unpredictability, and diverse security demands. This research proposes a novel hybrid optimization framework that integrates Reinforcement Learning (RL) with Multi-Objective Evolutionary Algorithms (MOEAs) to dynamically adapt scheduling decisions. The hybrid algorithm optimizes task execution time, enhances resource utilization, and strengthens security measures. Experimental evaluations against baseline optimization algorithms demonstrate the superiority of the proposed approach in achieving secure and efficient task scheduling for cloud-edge computing systems.