A FEDERATED AI MODEL FOR REAL-TIME THREAT DETECTION IN MULTI-CLOUD ENVIRONMENTS

The increase in the number of multi-cloud environments makes it difficult to achieve effective and robust cybersecurity due to unique challenges to the management of distributed and heterogeneous platforms.Traditional centralized threat-detection systems have latency problems, data privacy limitations, and scalability problems.The current paper proposes a Federated AI Model to perform Real-Time Threat Detection in Multi-Cloud Environments, which aims at alleviating these limitations by applying the principles of federated learning.This method allows several cloud nodes to jointly train a global threat-detection model without sharing raw data, protecting data sovereignty and privacy.Local network activity logs and security telemetry are run through each node to identify abnormal behavior through lightweight deep-learning architectures optimized to run on edge inference.The models created locally are securely aggregated using a central coordinator using differential privacy and secure multiparty computation methods.The suggested system is applicable to detecting the advanced persistent threats (APTs), malware, and zero-day attacks within various cloud environments (e.g., AWS, Azure, GCP) and be compliant with data-protection laws (e.g., GDPR and HIPAA).The model has been proven to be highly accurate in detection, low in the false-positive rate, and A Federated AI Model for Real-Time Threat Detection in Multi-Cloud Environments https://iaeme.com/Home/journal/IJAIRD73