PayShield AI — Protect Before You Pay

Inspiration

Online payments have made shopping and sending money easier, but they have also created opportunities for phishing, fake payment messages, fraudulent links, and impersonation scams.

A message that looks like it came from PayPal can convince someone to click a suspicious link or share sensitive information. I wanted to build a project that helps people pause, identify warning signs, and make safer decisions before making a payment.

That idea inspired PayShield AI — a payment safety platform built around one simple principle: protect people before they pay.

What it does

PayShield AI helps users assess suspicious payment-related messages and links.

Its key features include:

  • Message Risk Analyzer: Examines payment-related messages for suspicious language, urgency, credential requests, and possible phishing indicators.
  • Smart URL Inspector: Checks links for suspicious domains, PayPal impersonation patterns, and insecure HTTP connections.
  • Risk Assessment: Displays a risk score, risk category, and explanations of detected warning signs.
  • Security Analytics Dashboard: Tracks analysis results and displays high-, medium-, and low-risk counts.
  • Multilingual Support: Provides interface and warning translations in English, Hindi, Tamil, Telugu, and Kannada.
  • PDF Reports: Allows users to generate downloadable security assessment reports.
  • PayPal Sandbox Checkout: Demonstrates a separate test checkout flow through a backend connected to PayPal Sandbox.

How I built it

I developed PayShield AI as a web-based prototype using HTML, CSS, and JavaScript.

The frontend is hosted on GitHub Pages, while the backend is built with Node.js and Express and deployed on Render.

I implemented heuristic-based risk detection to identify suspicious patterns in messages and URLs. The results include risk scores and explanations so users can understand why a message or link may be suspicious.

I also integrated a PayPal Sandbox checkout flow. The backend securely handles order creation and payment capture, keeping PayPal credentials on the server rather than exposing them in the frontend.

The project also includes a security analytics dashboard, multilingual warnings, and PDF report generation.

Challenges I faced

Building this project involved several challenges:

  1. False positives: Some harmless messages contain words such as password or payment details. I refined the detection logic to distinguish warnings from messages that explicitly say sensitive information is not being requested.

  2. Backend deployment: Connecting the frontend to a Node.js backend required debugging deployment issues, environment variables, and API communication.

  3. PayPal integration: Configuring Sandbox credentials and resolving authentication errors took careful testing. I eventually got the demo checkout working.

  4. User experience: I wanted the risk results to be understandable, so I added clear explanations, analytics, translations, and downloadable reports.

  5. Responsible security messaging: A risk score cannot guarantee that a payment or website is safe. I made sure the application communicates that its results are assessments, not guarantees.

What I learned

This project helped me understand how frontend and backend systems work together, how to deploy a web application, how to integrate payment APIs, and how to approach basic cybersecurity detection.

I also learned that building a security tool is not just about detecting suspicious activity. It is equally important to explain the findings clearly and avoid giving users a false sense of security.

What's next

I would like to improve PayShield AI with more advanced AI-powered threat analysis, stronger URL reputation checks, improved detection accuracy, and additional language support.

The current version is a working prototype. Its heuristic assessments are not a substitute for independent verification, and its PayPal integration is configured for Sandbox testing rather than real payments.

PayShield AI — Trust before you pay. Protect before you pay.

Built With

Share this project:

Updates

Submission history