CyberShield EDU: An AI-Powered Platform for Detection and Prevention of Student-Targeted Online Frauds

With the rapid growth of digital communication, students are increasingly exposed to cyber threats such as phishing links, fake job offers, scholarship scams, and impersonation attacks. Traditional detection systems rely on static rules or keyword-based filtering, which are often ineffective against modern, context-aware attacks. This research proposes CyberShield-EDU, an AI-powered cybersecurity platform designed specifically for students. The system integrates Natural Language Processing (NLP), Optical Character Recognition (OCR), and heuristic-based URL analysis to detect threats across multiple input formats including text, images, PDFs, and URLs. The core of the system utilizes a DistilBERT model for contextual scam detection, supported by Shannon Entropy and Levenshtein Distance algorithms for identifying suspicious URLs. The platform also includes an educational module to improve user awareness and an admin dashboard for monitoring system activity. The proposed system provides a multi-layered detection approach, ensuring faster response time, improved accuracy, and enhanced user engagement while maintaining data privacy.