DESIGN AND EVALUATION OF AN AI-DRIVEN RESPONSIBLE COMPUTING FRAMEWORK FOR ETHICAL DECISION-MAKING IN CYBERSECURITY

Abstract

Artificial Intelligence (AI) has significantly improved cybersecurity through intelligent threat detection and automated response; however, many existing AI-based security systems operate as black boxes with limited transparency, accountability, and ethical governance. This study proposes an AI-Driven Responsible Computing Framework (AI-RCF) that integrates Artificial Intelligence, Explainable Artificial Intelligence (XAI), Responsible Computing principles, policy compliance, and continuous learning to support ethical decision-making in cybersecurity. The framework comprises modules for security data collection, AI-based threat intelligence, ethical risk assessment, explainable decision support, policy compliance verification, automated response, and continuous governance. The system was implemented using Python, TensorFlow, Spring Boot, React.js, PostgreSQL, and Docker, and evaluated using enterprise cybersecurity datasets. Experimental results demonstrate superior performance over conventional AI-based cybersecurity systems, achieving 98.4% accuracy, 98.1% precision, 97.8% recall, 98.0% F1-score, and an AUC of 0.991. Ethical evaluation also yielded high scores, including 97% ethical compliance, 95% explainability, 96% transparency, 94% fairness, 96% accountability, 98% privacy compliance, and 94% user trust. The findings indicate that integrating Responsible Computing principles enhances both technical performance and trustworthy AI decision-making without compromising detection effectiveness. The proposed AI-RCF provides a practical, scalable, and ethically compliant solution for securing enterprise, cloud, IoT, healthcare, financial, and critical infrastructure environments while promoting transparency, accountability, and regulatory compliance in AI-driven cybersecurity.