Work in Progress: Artificial Intelligence in Cybersecurity - Developing a Framework for Ethically Responsible Practices

Background: Artificial intelligence (AI) in cybersecurity is rapidly evolving, with the MANTRA research project aiming to enhance the security of critical infrastructures via a decentralized cybersecurity mesh. Despite the existence of regulatory frameworks such as the EU AI Act, ethical challenges specific to AI-driven cybersecurity remain inadequately addressed. Aim: This paper seeks to develop and validate a practical ethical framework that integrates core ethical principles like beneficence, justice, respect for autonomy and non-maleficence, to guide the design and implementation of AI-based cybersecurity solutions as developed in project MANTRA. Method: An iterative framework development process was employed, combining an extensive literature review, expert engagement (including three Ethical Board meetings) and the application of ethical risk management methods. The framework was structured using ethical risk matrices, the Delphi method and consensus-oriented procedures, with inputs drawn from ALTAI guidelines and other frameworks like the Ethical Impact Assessment. Outcome: The resulting ethical framework with its adaptable architecture systematically identifies, assesses and aims to alleviate ethical risks in AI-driven cybersecurity. Conclusion: This paper demonstrates how we developed tailored ethical framework for a cybersecurity AI project. By embedding ethical considerations directly into decisionmaking processes, the framework supports enhanced societal trust and operational integrity, addressing the unique challenges posed by decentralized, AI-powered cybersecurity. It promotes a proactive approach to integrating ethical considerations into AI-driven cybersecurity solutions, ensuring that technological advancements align with societal values and responsibilities.