GENESIS-AI Self -Improving Generative Intelligence for Cybersecurity in Complex Infrastructures
- Coordinator
- Sveuciliste U Zagrebu Fakultet Elektrotehnike I Racunarstva, Croatia
- Runs
- 1 October 2026 to 30 September 2029
- Status
- Running
- EU funding
- €13.7m
- Total cost
- €13.7m
- Partners
- 15 organisations in 10 countries
- Type
- Research and Innovation Action
- Grant agreement
- 101308552
What the project is doing
In the project's own words, from its summary on CORDIS.
GENESIS-AI develops a self-improving, self-adapting, and regulation-aware Generative AI ecosystem for cybersecurity, enabling Europe to defend its digital infrastructure with trustworthy, explainable, and privacy-preserving intelligence. Current security platforms rely on static, siloed analytics that fail against evolving multi-vector and zero-day threats. GENESIS-AI transforms this landscape through a reflexive, federated, and generative architecture that continuously learns, reasons, and governs itself.
At its core, GENESIS-AI integrates Multi-Agent Systems (MAS) and Mixture-of-Experts (MoE) paradigms within five interoperable Building Blocks:(1) Reflexive Governance and Interoperability Layer coordinating heterogeneous AI systems via a shared ontology and adaptive policies;(2) Knowledge, Compliance, and Explainability Intelligence Layer that translates EU regulations (AI Act, NIS2, GDPR) into executable rules;(3) Generative Cyber Defence and Adversarial Resilience Framework coupling generative models with adaptive Red/Blue Team co-evolution and autonomous self-healing;(4) Federated Privacy, Provenance, and Integrity Framework providing verifiable, GDPR-compliant learning and traceable model lineage; and(5) Efficiency, Sustainability, and Validation Stack combining quantisation-aware compression with large-scale validation in a Secure Cyber Range.
Through collaboration among Europe’s leading AI, cybersecurity, and industrial partners (Siemens, Airbus, Bosch), GENESIS-AI advances the state of the art from TRL 3-4 to TRL 6. It delivers a sovereign, trusted, and energy-efficient GenAI framework that autonomously defends, explains, and complies - turning cybersecurity from reactive protection into proactive, reflexive intelligence.
Topics
Who is involved
| Organisation | City | Country | Role |
|---|---|---|---|
| Sveuciliste U Zagrebu Fakultet Elektrotehnike I Racunarstva | Zagreb | Croatia | Coordinator |
| Technikon Forschungs- Und Planungsgesellschaft Mbh | Villach | Austria | Participant |
| Katholieke Universiteit Leuven | Leuven | Belgium | Participant |
| Robert Bosch Car Multimedia GmbH | Hildesheim | Germany | Participant |
| Siemens Aktiengesellschaft | Munchen | Germany | Participant |
| Technische Universitat Darmstadt | Darmstadt | Germany | Participant |
| Barcelona Supercomputing Center Centro Nacional De Supercomputacion | Barcelona | Spain | Participant |
| Ikerlan S. Coop | Mondragon | Spain | Participant |
| Turun Yliopisto | Turku | Finland | Participant |
| Airbus Protect | Blagnac | France | Participant |
| SecureML jednostavno drustvo s ogranicenom odgovornoscu za racunalno programiranje | Sesvete | Croatia | Participant |
| Budapesti Muszaki Es Gazdasagtudomanyi Egyetem | Budapest | Hungary | Participant |
| Universita Degli Studi Di Pavia | Pavia | Italy | Participant |
| The Chancellor, Masters and Scholars of the University of Oxford | Oxford | United Kingdom | Participant |
| University of Essex | Colchester | United Kingdom | Participant |
Source: CORDIS, European Commission. Data of 5 Nov 2026, reused under the Commission's reuse policy. View on CORDIS · Grant DOI