metadata

license: apache-2.0
title: PhishGuardian AI
sdk: gradio
emoji: ⚡
colorFrom: red
colorTo: yellow
pinned: false
thumbnail: >-
  https://cdn-uploads.huggingface.co/production/uploads/6838c79ebfdc8508f8fdead2/Y8nN6RWVuCMaJNmLG5X-A.png
short_description: UDSM AI-powered tool for real-time phishing email detection.

PhishGuardian AI 🛡️

AI-powered phishing email detection using DistilBERT for real-time security analysis.

Overview

PhishGuardian AI is an intelligent email security tool that classifies emails as legitimate or phishing using a fine-tuned DistilBERT model. Built for the University of Dar es Salaam (UDSM) community, it provides instant threat assessment through an intuitive web interface.

Features

Real-time Detection: Instant email classification with confidence scoring
Advanced AI Model: Fine-tuned DistilBERT (cybersectony/phishing-email-detection-distilbert_v2.4.1)
User-friendly Interface: Clean Gradio web interface with visual risk indicators
Comprehensive Analysis: Detailed probability breakdown for all threat categories
Educational Tool: Built-in examples and security recommendations

Quick Start

Online Access

Visit the deployed Space: https://huggingface.co/spaces/MUFASA25/phishguardian-ai

Local Development

git clone https://huggingface.co/spaces/MUFASA25/phishguardian-ai
cd phishguardian-ai
pip install -r requirements.txt
python app.py

Usage

Input: Paste email content into the text area
Analyze: Click "Analyze Email" for instant results
Review: Examine risk level, confidence score, and detailed analysis
Act: Follow provided security recommendations

Example Analysis

Input: Suspicious email with urgent account verification request

Output:

🚨 HIGH RISK
Primary Classification: Phishing Email
Confidence: 92.8%
Recommendation: Do not click any links or provide personal information.

Technical Specifications

Model: DistilBERT-base fine-tuned for sequence classification
Input Limit: 512 tokens
Classes: Legitimate Email, Phishing Email, Suspicious Content, Other
Framework: Transformers, PyTorch, Gradio
Deployment: Hugging Face Spaces

Requirements

gradio>=4.0.0
transformers>=4.21.0
torch>=1.12.0

Contributing

Fork the repository
Create feature branch (git checkout -b feature/enhancement)
Commit changes (git commit -m 'Add enhancement')
Push to branch (git push origin feature/enhancement)
Open Pull Request

License

Licensed under Apache 2.0. See LICENSE for details.

Contact

Developer: MUFASA25
Email: [email protected]
Institution: University of Dar es Salaam (UDSM)
Profile: https://huggingface.co/MUFASA25

⚠️ Disclaimer: This tool is for educational and awareness purposes. Always follow your organization's security protocols and use professional judgment when handling suspicious emails.