GENERATIVE AI–DRIVEN SITUATIONAL AWARENESS FOR NATIONAL SECURITY: MULTI-MODAL DATA FUSION, ANOMALY DETECTION, AND AUTONOMOUS THREAT RESPONSE

Ensuring real-time situational awareness in national security operations is increasingly challenging due to the growing complexity of multi-domain threats and the diversity of intelligence sources. Existing analytical and centralized fusion approaches struggle with data scarcity, privacy constraints, and limited adaptability to evolving scenarios. To address these challenges, this study proposes a novel Generative AI–driven situational awareness framework that integrates multi-modal data fusion, generative anomaly simulation, and autonomous response optimization within a unified architecture. The framework leverages diffusion models and Generative Adversarial Networks (GANs) to generate high-fidelity synthetic anomalies, enabling robust detection of rare events. A federated multimodal fusion mechanism ensures secure cross-agency collaboration without exchanging raw data, while a reinforcement learning (RL) and multi-agent RL (MARL)-based decision intelligence module delivers machine-speed response strategies. Furthermore, explainable AI (XAI) and conformal uncertainty quantification enhance system transparency and trustworthiness. Extensive experiments are conducted on benchmark datasets, including xView and xBD for geospatial intelligence and UNSW-NB15 and CIC-IDS2018 for cybersecurity. The results demonstrate that the proposed framework significantly outperforms baseline methods, improving F1-scores by up to 10.6%, achieving 15.4% higher fusion accuracy, and reducing response latency by 42.9%. Additionally, XAI-based explanations reduce analyst verification time by 23%, ensuring compliance with the NIST AI Risk Management Framework. These findings confirm that the proposed approach provides a scalable, trustworthy, and adaptive solution for enhancing real-time situational awareness in mission-critical national security environments.