About

GENESIS-AI is a next-generation European initiative developing a self-improving, regulation-aware Generative AI ecosystem for cybersecurity. The project combines Multi-Agent Systems (MAS), Mixture-of-Experts (MoE), federated learning, and explainable AI to create a trusted cyber defence framework that continuously adapts to emerging threats while remaining aligned with European regulations such as the AI Act, NIS2, and GDPR. Rather than relying on static security solutions, GENESIS-AI enables autonomous reasoning, adaptive protection, compliance intelligence, and self-healing capabilities across cloud, edge, and critical infrastructure environments. By integrating governance, transparency, privacy, and resilience into a unified architecture, GENESIS-AI aims to strengthen Europe's digital sovereignty and transform cybersecurity from reactive defence into proactive, trustworthy intelligence.

Vision

GENESIS-AI envisions a future in which cybersecurity systems are not merely reactive tools, but autonomous, trustworthy, and continuously evolving digital guardians. Our vision is to establish a sovereign European AI ecosystem that can understand, anticipate, and counter emerging cyber threats while remaining fully aligned with European values, regulations, and ethical principles. By combining self-improving Generative AI, multi-agent intelligence, explainability, privacy-preserving learning, and adaptive governance, GENESIS-AI aims to transform cybersecurity into a proactive and resilient capability. The project seeks to empower organisations across critical sectors with secure, transparent, and regulation-aware AI technologies, strengthening Europe's digital sovereignty and setting new global standards for trustworthy AI-driven cyber defence.

Keyfacts

Reference

Reference Number:
101308552
Programme type:
Horizon Europe
Programme acronym:
HORIZON-CL3-2025-02-CS-ECCC-01

Duration

Project Start: 01.10.2026
36 Months

Cost and Funding

Costs:
€ 13.709.654,99
Funding:
€ 13.709.654,99

Motivation

Europe’s digital infrastructure is facing an increasingly complex threat landscape, where cyberattacks evolve faster than traditional security solutions can respond. Current AI-powered cybersecurity tools are often fragmented, opaque, and dependent on static intelligence, making them insufficient against sophisticated, multi-vector, and zero-day threats. At the same time, organisations must navigate a growing regulatory landscape shaped by the AI Act, NIS2, GDPR, and the Cyber Resilience Act. GENESIS-AI was created to address these challenges by building a trustworthy, explainable, and regulation-aware Generative AI ecosystem. By combining adaptive intelligence, privacy-preserving learning, autonomous resilience, and compliance-by-design, the project aims to strengthen Europe’s cyber resilience, digital sovereignty, and trust in AI-driven cybersecurity solutions.

Mission and Objectives

GENESIS-AI’s mission is to transform cybersecurity from a reactive discipline into a proactive, trustworthy, and continuously evolving intelligence ecosystem. By combining Generative AI, multi-agent collaboration, federated learning, and compliance-by-design, the project seeks to strengthen Europe’s digital sovereignty while ensuring transparency, resilience, and alignment with European regulations. GENESIS-AI will deliver a new generation of cybersecurity solutions capable of learning, adapting, explaining, and governing themselves in rapidly changing threat environments.

Key Objectives

  • O1. Establish a Governance and Interoperability Framework for Heterogeneous AI Systems
  • O2. Enable Reflexive, Self-Improving Multi-Agent Governance
  • O3. Operationalise Knowledge and Compliance Intelligence
  • O4. Deploy Generative Cybersecurity Experts
  • O5. Reinforce Adversarial Robustness through Adaptive Red/Blue Teaming
  • O6. Ensure Privacy-Preserving Distributed Learning
  • O7. Achieve Explainability, Fairness, and Transparency in Decision-Making
  • O8. Guarantee Integrity and Provenance Across the AI Lifecycle
  • O9. Develop Autonomous Self-Healing and Resilient AI Systems
  • O10. Validate and Demonstrate the Integrated Ecosystem in a Secure Cyber Range

Work Packages

WP1

Project, Risk, and Innovation Management
3%
01.10.2026
30.09.2029
WP1 is responsible for the operational management and technical vitality of the project, encompassing man-agement tasks on contractual, financial, legal, technical, administrative and ethical levels. The main objective is the establishment of a sound and flexible project management structure and an effective risk management strategy to avoid and properly react to deviations from the work plan. The Coordinator UNIZG-FER will be supported by the project management experts from TEC. Related to all Main Objectives.

WP2

Governance and Interoperability Framework for Generative AI Cybersecurity
4%
01.10.2026
30.11.2028
Develop the self-improving control and communication backbone of the GenAI cybersecurity ecosystem, a network of multi-agent modules that coordinate experts, enforce policies, authenticate components, and con-tinuously optimise performance and compliance. WP2 ensures adaptive governance: it observes perfor-mance metrics, adjusts model cooperation strategies, and drives self-improvement loops across all WPs. Main components: Judge, Analyser, Optimiser, Regulator, Attacker, Tester, Authentication & Access Control, SBOM/Provenance, Adapter, Broker, Translator, Harmoniser, Constraint Guard, Feedback Collator. Related to Main Objectives: O1, O2, O3, O7, O8.

WP3

Knowledge and Compliance Intelligence Layer
0%
01.12.2026
31.01.2029
Develop specialized MOE-based Global Experts that continuously refine their knowledge of regulations, security requirements, and interoperability standards. Provide self-improving domain intelligence, updating itself as new regulations, vulnerabilities, or best practices emerge. Main components: Regulation Expert, Se-curity Requirements Expert, Specification & Interoperability Expert. Related to Main Objectives: O1, O3, O7, O8.

WP4

Generative AI Cyber Protection and Cyber Operations Experts
0%
01.01.2027
31.03.2029
WP4 develops a suite of self-adapting, generative AI cybersecurity experts that autonomously detect, ana-lyse, and respond to advanced threats. Each expert targets a critical security domain - anomaly detection, malware analysis, security fuzzing, incident response, and threat intelligence correlation - and collectively forms a resilient, dynamic protection ecosystem. By integrating feedback from the Governance MAS (WP2), Adaptive Regulation Expert (WP3), and the self-learning Federated Learning (FL) Coordinator, these agents will continuously learn, adapt, and refine their strategies through real-world and simulated feedback. This ensures proactive security and sustained effectiveness against evolving threats. WP4 also focuses on explain-ability and human oversight, generating interpretable insights to help analysts understand automated deci-sions and strengthen trust in AI-driven operations. Related to Main Objectives: O2, O4, O5, O8.

WP5

Trustworthy, Robust, and Secure AI Experts
0%
01.03.2027
31.05.2029
Objectives: Develop a self-improving and self-adapting AI security layer to ensure models remain robust, private, and trustworthy even as new attacks, datasets, and regulations emerge. WP5 protects the GenAI ecosystem itself, embedding self-healing and adaptive privacy-preserving mechanisms. Main components include Red/Blue Team Agents, Privacy & FL Coordinator, Bias & Explainability Agent, Authentication & Access Control MAS, Model Fingerprinting & Provenance. Related to Main Objectives: O5, O6, O7, O8.

WP6

Secure Cyber Range & Sector Pilots
0%
01.09.2027
30.09.2029
Demonstrate the feasibility of GENESIS-AI methods and tools through four real-world demonstrators, ad-dressing various safety and security challenges at both HW and SW levels. Implement realistic scenarios to assess the effectiveness and scalability of the solutions. Evaluate the benefits and effectiveness of the ap-proach, using results and KPIs. Shape industry-wide best practices in AI for security systems. Related to Main Objectives: O2, O4, O5, O7, O9, O10.

WP7

Dissemination, Communication, Exploitation, and Standardisation
3%
01.10.2026
30.09.2029
WP7 focuses on maximizing the project’s impact and long-term success through a comprehensive communi-cation and dissemination strategy, sharing project goals and results with relevant communities. It supports the EU’s scientific research objectives, helps translate scientific discoveries into practical benefits, and fosters innovative applications based on research findings. WP7 also ensures proper management of intellectual property and contributes to standardisation efforts, advancing the project beyond the current state-of-the-art.

Follow us on