Federated Sovereign AI Ecosystems: A Framework for Trusted Governance and Cross-Border Intelligence Interoperability

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The use of Artificial Intelligence (AI) in Cybersecurity and cross-border intelligence sharing has increased in relevance and importance in recent years; however, current Centralised AI architectures are plagued with data privacy, Digital sovereignty, Governance and regulatory compliance concerns. This study introduces a Federated Sovereign AI Ecosystem (FSAIE) designed to foster the sharing of intelligence within a federated learning framework, securely integrate data interoperability, and ensure national data sovereignty and trusted Governance. Testing of the framework was conducted on the UNSW-NB15 cybersecurity dataset that consists of 257,673 network traffic records with 49 features, and the World Governance Indicators (WGI) dataset. Four models (Random Forest, XGBoost, Deep Neural Network (DNN), Federated Learning) were built and evaluated for accuracy, precision, recall, F1-score, ROC-AUC, trust index, interoperability and communication cost. The experimental results show that the XGBoost model has the highest “accuracy (99.05%), precision (99.50%), recall (99.01%), F1 score (99.25%) and ROC-AUC (0.9995%)” than the Random Forest model (accuracy: 98.08%) and DNN model (accuracy: 97.18%). The proposed ecosystem also showed an impressive Trust Index of 94.56 and achieved a 100% efficiency in intelligence exchange and a communication cost of 21,096.22 KB per round, which establishes a secure and efficient cross-border collaboration. The proposed FSAIE is a scalable, privacy-first, and governance-enlightened solution closely balancing predictive performance with trusted interoperability and national data sovereignty to make collaborative AI-driven cybersecurity applications a reality.