AINeutralarXiv – CS AI · 7h ago6/10
🧠
Agentic AI for Cybersecurity: A Meta-Cognitive Architecture for Governable Autonomy
Researchers propose a meta-cognitive agentic AI framework for cybersecurity that replaces deterministic SOAR systems with probabilistic decision-making agents coordinated through uncertainty evaluation. Empirical testing on benchmark datasets demonstrates improved robustness, lower false positives, and better-calibrated confidence estimates compared to traditional approaches.