- Version
- Download 0
- File Size 339.98 KB
- File Count 1
- Create Date June 26, 2026
- Last Updated June 26, 2026
AI-Driven Cyber Threat Discovery and Risk Prediction: Methods, Models, And Ethical Dimensions
ABSTRACT
The growing complexity of cyber threats has outpaced traditional security tools, driving a shift toward AI-powered cybersecurity solutions. This study explores recent advances in machine learning and deep learning for threat detection, focusing on automation and predictive analytics. Using a PRISMA-guided review, 10 peer-reviewed studies from 2023 were analyzed. A conceptual taxonomy was developed to classify AI techniques by learning type, function, and transparency. Predictive models—Random Forest, graph-based models, and reinforcement learning—were tested, showing up to 95.95% accuracy, especially when socio-demographic data were included. Findings emphasize the power of explainable, context-aware AI in early threat detection. However, challenges remain in ethical deployment, dual-use risks, and readiness of subject matter experts. The study proposes a framework for scalable, interpretable AI in cybersecurity and recommends further validation and policy integration. AI is a critical enabler of proactive threat response, but its use must align with ethical, technical, and governance best practices..
Keywords: Artificial Intelligence, Cybersecurity, Threat Discovery, Machine Learning, Explainable AI, Advanced Persistent Threats, Predictive Analytics, Bio-Cybersecurity