🛡️ PhishShield – Phishing Takedown Platform for PSBs 💡 Inspiration The increasing cases of phishing scams targeting customers of Public Sector Banks (PSBs) in India—especially websites impersonating official portals—motivated me to create a solution. Many such phishing domains stay live for days due to lack of quick detection, manual takedown processes, and jurisdictional hurdles. A recent analysis of takedown delays and the lack of automation in this space revealed a major gap. I envisioned PhishShield, a platform that uses AI and automation to enable faster, smarter takedowns, starting with PSBs.
⚙️ What It Does PhishShield is an AI-powered phishing takedown platform that:
Detects phishing websites using ML models trained on domain and content patterns.
Allows banks, CERTs, and users to report phishing links via a reporting portal.
Automates takedown requests by integrating with domain registrars and escalates through CERTs if needed.
Tracks progress in a centralized dashboard with real-time status updates.
Maintains secure, immutable records of actions taken using blockchain.
Ensures data security with encryption and identity/authentication services.
🏗️ How I Built It The platform is hosted on IBM Cloud and consists of multiple integrated modules:
- AI-Powered Phishing Detection Tools: IBM Watson Studio, Watson NLU
ML Models: Trained using Scikit-learn and TensorFlow
Serverless detection logic with IBM Cloud Functions
- Reporting Interface Frontend: Built using React.js
Backend: Python Flask APIs via IBM API Connect
Allows banks, users, and CERTs to report suspicious URLs
- Centralized Dashboard Tools: IBM Cloud Foundry, IBM Cloud Databases
Tech Stack: React.js + Flask backend
Tracks reported phishing sites, takedown status, timestamps
- Registrar Integration Takedown automation via REST APIs (GoDaddy, Namecheap, etc.)
Escalation process built into workflow for non-API registrars
- Blockchain-Backed Action Logs Tools: IBM Blockchain Platform
Secure and immutable record of takedown requests and status
- Real-Time Progress Tracking IBM Cloud Monitoring + WebSocket polling
Updates visible to all stakeholders instantly
- Security & Compliance Tools: IBM Cloud Identity, Encrypted IBM Cloud Databases
Auth: OAuth 2.0 and JWT-based authentication for all users
🧗 Challenges I Ran Into Acquiring and cleaning phishing domain datasets.
Ensuring accurate detection while minimizing false positives.
Automating registrar takedown requests where API support is limited.
Integrating IBM Blockchain with Flask securely.
Balancing performance with free-tier scalability on IBM Cloud.
🏅 Accomplishments That I'm Proud Of Building a fully functional prototype that mimics real-world phishing takedown operations.
Automating takedown requests, drastically reducing manual effort.
Creating a real-time dashboard for multiple stakeholders.
Successfully integrating multiple IBM Cloud services, AI tools, and Blockchain.
📚 What I Learned End-to-end development using the IBM Cloud ecosystem.
Working with Watson Studio and NLU for NLP-based threat detection.
Implementing secure APIs, encryption standards, and OAuth.
Practical exposure to cross-border cyber law considerations.
🔮 What's Next for PhishShield Training the AI model with larger, real-world datasets.
Adding multilingual support for phishing content in regional Indian languages.
Partnering with PSBs and government agencies to deploy a live pilot.
Introducing alerting systems via SMS/Email for banks/customers.
Integrating with Indian CERT and law enforcement APIs for real-time action.
Built With
- ibm
- python
- streamlit
Log in or sign up for Devpost to join the conversation.