AIAgent4Cyber
Autonomous AI Agents for Proactive Cyber Defence
Autonomous AI agents for proactive cyber defence, combining explainable AI, reinforcement learning, and federated learning across a 12-partner European consortium.
- Programme
- Digital Europe Programme
- Grant
- €433,992
- Period
- 36 months
Problem
Modern cyber threats evolve faster than human analysts can respond. Security operations centres face an increasing volume and sophistication of attacks, while alert fatigue and skills shortages limit the effectiveness of manual detection and response workflows.
Existing automated tools rely on static signatures and centralised architectures that struggle with novel attack patterns, cross-organisation threat correlation, and the need to justify decisions to human operators.
Requirement: Intelligent agents that can detect, reason about, and respond to threats autonomously while remaining explainable, privacy-preserving, and adaptable across heterogeneous network environments.
Approach
AIAgent4Cyber develops and deploys AI-driven autonomous agents for proactive cyber defence across a pan-European network of partners. The project targets key technology deployment in real operational environments.
Cognifinity Contributions
- Explainability (XAI) Modules: interpretability components that surface reasoning behind agent decisions to security analysts, supporting auditability and trust.
- RL-Based Agent Coordination: reinforcement learning policies for multi-agent coordination, enabling adaptive response strategies across distributed network environments.
- Anomaly Detection and Log Analysis: AI-driven analysis of security event streams and logs to surface high-fidelity alerts with reduced false positive rates.
- Federated Learning Architecture: privacy-preserving collaborative training that allows partners to improve shared models without sharing raw network data across organisational boundaries.
Consortium
The project brings together 12 partners from seven countries: Spain, Austria, Finland, Latvia, Luxembourg, Greece, France, and Lithuania. The consortium spans research institutions, technology providers, and operational cybersecurity organisations. The kick-off meeting took place in Vienna on 29 January 2026 at SBA Research.
Interested in AI-Driven Cyber Defence?
Get in touch to learn more about our work in autonomous cybersecurity agents and explainable AI for security operations.
Contact R&D Team