The Project Overview Traditional rule-based phishing detection mechanisms struggle to keep up with modern cyber threats. To solve this, I engineered PhishX, an intelligent, AI-driven sandbox capable of analyzing and classifying malicious URLs in real-time.

How the AI Works The uploaded screenshots demonstrate how the framework handles different real-world security scenarios dynamically:

Scenario 1 (Green): Verifying legitimate URLs and matching them against safe profiles.

Scenario 2 (Red): Intercepting active phishing vectors with a high risk-score (92.6%).

Scenario 3 (Yellow): Resiliently handling dead or offline links by predicting intent purely through URL structure.

Explainable AI (XAI) Integration Instead of acting as a "black box," PhishX integrates SHAP (SHapley Additive exPlanations) metrics. This ensures absolute transparency by displaying exactly which features—such as lines of code, URL similarity index, or external references—contributed to the final security classification.