AdaPhish: AI-Powered Adaptive Defense and Education Resource Against Deceptive Emails
Project Overview
The document discusses the integration of generative AI in education, focusing on its application in enhancing cybersecurity awareness through the AdaPhish platform. This AI-powered tool employs large language models and vector databases to detect and analyze phishing emails, automating the anonymization of sensitive data and offering real-time alerts and trend analysis. By addressing the limitations of traditional phishing detection methods, AdaPhish not only improves the accuracy of identifying phishing attempts but also fosters a collaborative learning environment through its phish bowl system, where users can contribute known phishing cases for communal analysis. The findings suggest that such generative AI applications significantly enhance cybersecurity education by providing practical, hands-on learning experiences, ultimately leading to improved awareness and resilience against phishing threats in educational settings.
Key Applications
AdaPhish - an AI-powered platform for phishing detection and education
Context: Cybersecurity education for organizations and individuals
Implementation: Uses LLMs for email anonymization and analysis, employing a collaborative phish bowl for tracking phishing trends and real-time alerts
Outcomes: Improved detection accuracy, real-time adaptability, and effective educational tools for understanding phishing threats
Challenges: Centralized architecture risks scalability and insider threats, potential for data loss
Implementation Barriers
Technical Barrier
Centralized architecture can become a bottleneck as email volume grows, risking scalability issues. Additionally, it increases exposure to insider threats.
Proposed Solutions: Future enhancements could involve decentralized storage solutions, strengthening access controls, implementing audit trails, and exploring decentralized verification methods.
Project Team
Rei Meguro
Researcher
Ng S. T. Chong
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Rei Meguro, Ng S. T. Chong
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai