PhishGuard AI

An AI-powered shield designed to protect elderly users from phishing scams and malicious links.

Comment 1

Inspiration

My inspiration comes from seeing how vulnerable elderly people are to digital scams. In my community, I noticed that many senior citizens fall victim to phishing links because they lack tech experience. As a Computer Science student interested in Cybersecurity, I wanted to build a simple, accessible tool that acts as a first line of defense for them.

What it does

PhishGuard AI is a user-friendly platform where users can paste suspicious messages or links. The AI analyzes the text and patterns to determine if it's a phishing attempt. It then provides a clear, color-coded verdict: "Safe" (Green) or "Dangerous" (Red), specifically designed with a simple interface for non-technical users.

How I built it

I designed the frontend using HTML and CSS to ensure it is clean and accessible. For the "brain" of the project, I used Baidu's MeDo AI platform to train a Natural Language Processing (NLP) model. I used Python to bridge the gap between the user input and the AI analysis.

Challenges I ran into

As a first-year student, one of the biggest challenges was understanding how to integrate an AI model into a web interface. Also, tailoring the AI to recognize local scam patterns required careful training of the data.

Accomplishments that we're proud of

I am proud of creating a project that has a real social impact. Building a functional bridge between complex AI technology and an easy-to-use interface for elderly people is a major milestone for me as a beginner

What we learned

I learned how to use Low-Code AI platforms like MeDo to solve security problems. I also improved my skills in structuring a project from scratch and thinking about the user experience (UX) for people who are not experts in technology

What's next for PhishGuard AI

The next step is to add support for more languages, including Arabic and Moroccan Darija, and to develop a mobile app version to provide real-time protection on smartphones

Built With

Share this project:

Updates

posted an update —

As I move closer to the final demonstration, my current focus is on:

  • Dataset Optimization: Fine-tuning the machine learning model with more diverse phishing samples to reduce false positives.
  • ⁠XAI Visualization: Improving how the "Explainable" part of the AI displays logic to the user, making security insights easier to understand.
  • ⁠Final Presentation Prep: Preparing the technical documentation and demo video for the upcoming showcase at ENSA Marrakech. The goal is to provide a robust, transparent, and user-friendly security tool. Stay tuned!

Log in or sign up for Devpost to join the conversation.

posted an update —

Core Technical Features

  • Real-time URL Analysis: Leveraging machine learning to scan for suspicious patterns and domain masquerading.
  • ⁠XAI Dashboard: Using Explainable AI to provide users with a "Safety Score" and the reasons behind it.
  • ⁠Secure Tech Stack: Developed with a robust Python backend and a responsive frontend for seamless interaction.
  • ⁠Scalability: Designed to handle multiple requests efficiently, ensuring fast response times for threat detection.

Log in or sign up for Devpost to join the conversation.

posted an update —

PhishGuard AI: Transitioning from Development to a Functional Prototype. I have made significant progress in refining the core architecture of PhishGuard AI. The focus has been on integrating Explainable AI (XAI) to ensure that our phishing detection isn't just accurate, but also transparent for the end-user. Currently, the backend, powered by Python, is successfully analyzing URL patterns and metadata in real-time, providing clear justifications for its security ratings. I’ve also enhanced the user interface to ensure a seamless experience. As I prepare for the upcoming AISEC 2026 forum, I am finalizing the prototype to demonstrate how this solution effectively mitigates cyber threats while maintaining high performance and reliability.

Log in or sign up for Devpost to join the conversation.

Submission history