6th Generation (6G) networks is being enabled by network slicing to support services with varied performance requirements over shared physical infrastructure, in many respects, and even with logical separation, shared resources and interdependent network functions can (to some degree) expose individual slices to attacks and allow their effects to broaden to other slices, on the whole. Suspicious activity can be spotted by Artificial Intelligence (AI)-based threat detection, but mitigation decisions also are needing to account for prediction uncertainty and the service disruption caused by defensive actions. Estimated attack probability and prediction uncertainty with slice-level indicators, including traffic behaviour, resource utilization, latency and inter-slice reliances are being combined by this model, and is employing these inputs to assess security risk and select among graduated responses, namely monitoring, rate limiting, partial isolation, and full isolation. A Quality of Service (QoS)-constrained decision framework for easing direct and cross-slice attacks in 6G network slicing is being suggested by this study. The decision model studies residual attack exposure, QoS degradation, service availability, and mitigation overhead while enforcing slice-specific service constraints; this method will be assessed in a copied multi-slice environment using public intrusion-detection datasets to get attack and gentle traffic patterns, together with made slice-level resource conditions and direct and cross-slice attack scenarios. Its performance will be checked with fixed-threshold and adaptive mitigation baselines using detection quality, attack containment, false-isolation rate, isolation delay, latency, output, Service-Level Agreement (SLA) violations, and control overhead; the check will analyse whether representing prediction uncertainty can support more proportionate mitigation while upholding network service continuity.

Keywords- 6G Network Slicing; Intrusion Detection; Prediction Uncertainty; Quality of Service (QoS); Adaptive Mitigation; Cross-Slice Attacks; Network Security.