GENERATIVE AI POWERED

Detect Phishing
Before It Strikes

Advanced NLP analysis + AI explanation engine to protect you from social engineering attacks. Paste any suspicious message below.

3.2M+

Messages Scanned

94.7%

Detection Accuracy

12ms

Avg. Response Time

$ MESSAGE ANALYSIS TERMINAL

0 chars

ATTACK SIMULATION LAB

Phishing Attack
Simulator

See how attackers craft deceptive messages. Educational tool to recognize manipulation tactics before falling victim.

SELECT ATTACK SCENARIO

🏦

Bank Account Verification

Financial phishing via spoofed bank alerts

💼

Fake Job Offer

Employment scam targeting job seekers

🔐

OTP Theft Attack

Social engineering to steal one-time passwords

📦

Parcel Delivery Scam

Fake courier notifications with malicious links

🏆

Prize / Lottery Win

Too-good-to-be-true reward phishing

🏛️

Government Impersonation

Fake tax authority or legal threat scams

PROJECT OVERVIEW

About PhishGuard AI

🎯

The Problem

Phishing attacks cost organizations billions annually. Social engineering exploits human psychology, making technical defenses insufficient. Users need awareness training, not just filters.

🧠

How AI Is Used

Rule-based NLP detects linguistic patterns, urgency triggers, and suspicious domains. Claude AI generates personalized explanations and educational simulations that adapt to each message context.

🛡

Detection Engine

Multi-layer analysis: keyword scoring, domain validation, impersonation pattern matching, sentiment analysis for urgency/fear tactics, and link structure inspection.

⚡

Tech Stack

Frontend: HTML5, CSS3, Vanilla JS with cybersecurity dashboard UI. Backend: Python Flask API. AI: Claude (claude-sonnet-4-20250514) for natural language explanation and simulation generation.

HOW TO RUN LOCALLY

# 1. Install Python dependencies
pip install flask flask-cors anthropic

# 2. Set your Anthropic API key
export ANTHROPIC_API_KEY="your_key_here"

# 3. Start the Flask backend
cd backend
python app.py

# 4. Open the frontend
open frontend/index.html
# or serve it with: python -m http.server 8080

Detect Phishing Before It Strikes