Industry 4.0
Vol. 11 (2026), Issue 1, pg(s) 3-7

TECHNOLOGICAL BASIS OF “INDUSTRY 4.0”

AI-Enhanced Cybersecurity in Critical Infrastructures: A TRITON Framework and Review

  • 1 Technical University Sofia
  • 2 UBITECH
  • 3 EXUS
  • 4 Universidad Politécnica de Madrid
  • 5 K3Y
  • 6 Unversity of Genoa

Abstract

Artificial intelligence (AI) is reshaping the field of penetration testing by enabling faster, more adaptive simulations of cyber threats. This paper explores how AI can be ethically integrated into penetration testing processes, focusing on the European Defence Fundbacked TRITON project. TRITON proposes a comprehensive AI-driven framework for testing the security of military and critical infrastructure systems, combining technologies like machine learning, generative models, and reinforcement learning. The TRITON project reviews recent advancements in AI-supported pentesting and discusses how these tools can improve vulnerability detection, attack simulations, and threat modeling. Alongside the technical discussion, which is examined in the TRITON project, ethical concerns—including transparency, human oversight, and dual-use risks—are of the essence and must be addressed to ensure responsible use. Comparisons with other EU cybersecurity initiatives, such as AI4CYBER and CyberSecDome, highlight TRITON’s unique contributions and focus areas. Ultimately, it argues that AI-enhanced penetration testing can significantly strengthen cybersecurity defenses when implemented with appropriate safeguards and ethical oversight.

Keywords

References

  1. Al-Sinani, H. S., & Mitchell, C. J. (2025). PenTest++: Elevating Ethical Hacking with AI and Automation. arXiv preprint arXiv:2502.09484. https://arxiv.org/abs/2502.09484
  2. Ariadna Moreno, A., Sánchez, J., & Luna, J. (2025). Analysis of Autonomous Penetration Testing Through Reinforcement Learning and Recommender Systems. Sensors, 25(1), 211. https://doi.org/10.3390/s25010211
  3. European Commission. (2022). Trustworthy Artificial Intelligence for Cybersecurity Reinforcement and System Resilience (AI4CYBER). CORDIS Project ID 101070450. https://cordis.europa.eu/project/id/101070450
  4. TRITON Consortium. (2025). Generative Automation of Security Penetration Tests – Project Website. https://triton-edf.eu
  5. UBITECH. (2024, December 4). UBITECH Hosts Kick-off of TRITON EDF Action Project to Revolutionize Automated Security Penetration Testing. https://ubitech.eu
  6. Jabir, R., Le, J., Nguyen, C., Öner, U., & Ashour, M. (2024). Phishing Attacks in the Age of Generative Artificial Intelligence: A Systematic Review of Human Factors. AI, 6(8), 174. https://doi.org/10.3390/ai6080174
  7. TRITON Project Consortium. TRITON Project Partners Overview. Available online: https://triton-edf.eu/triton-consortium/ Ghanem, A., Ali, H., & Wang, Y. (2023). Hierarchical Reinforcement Learning for AI-Driven Cybersecurity Attack Path Discovery. Computers & Security, 130, 102868. https://doi.org/10.1016/j.cose.2023.102868
  8. Fernandes, D. A., Rodrigues, J. J., & Kumar, N. (2024). Using Reinforcement Learning to Simulate Advanced Persistent Threats. Journal of Network and Computer Applications, 228, 103487. https://doi.org/10.1016/j.jnca.2023.103487
  9. Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde- Farley, D., Ozair, S., ... & Bengio, Y. (2014). Generative Adversarial Nets. Advances in Neural Information Processing Systems, 27, 2672–2680.
  10. Doshi-Velez, F., & Kim, B. (2017). Towards a Rigorous Science of Interpretable Machine Learning. arXiv preprint arXiv:1702.08608. https://arxiv.org/abs/1702.08608
  11. ENISA. (2021). Artificial Intelligence Threat Landscape. European Union Agency for Cybersecurity. https://www.enisa.europa.eu/publications/artificial-intelligence-threat-landscape
  12. High-Level Expert Group on AI. (2019). Ethics Guidelines for Trustworthy AI. European Commission. https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai
  13. Shrestha, R., Mahmood, A., & Hu, J. (2021). Review of Deep Learning Algorithms and Architectures for Cybersecurity Applications. IEEE Access, 9, 29684–29713. https://doi.org/10.1109/ACCESS.2021.3058794
  14. Sommer, R., & Paxson, V. (2010). Outside the Closed World: On Using Machine Learning for Network Intrusion Detection. IEEE Symposium on Security and Privacy, 305–316. https://doi.org/10.1109/SP.2010.25
  15. CyberSecDome Consortium. (2025). CyberSecDome Project Overview and News Release. EIT Digital. https://28digital.eu/news/cybersecdome-eit-cybersecurity-platform
  16. Deng, G., Liu, Y., Mayoral-Vilches, V., Liu, P., Li, Y., Xu, Y., Zhang, T., Liu, Y., Pinzger, M., & Rass, S. (2024). PentestGPT: An LLM-empowered Automatic Penetration Testing Tool. https://arxiv.org/abs/2308.06782
  17. Shen, X., Wang, L., Li, Z., Chen, Y., Zhao, W., Sun, D., Wang, J., & Ruan, W. (2025). PentestAgent: Incorporating LLM Agents to Automated Penetration Testing. https://arxiv.org/abs/2411.05185

Article full text

Download PDF