PhishGuard AI: Fortifying Your Digital Frontier

Results are shown
The heatmap
Also can paste files
Dashboard and Front page
The text box for pasting any url, text and all

Inspiration

Phishing attacks and online scams are increasing rapidly, especially targeting students, elderly users, and online banking users. Many people cannot easily identify fake messages, malicious links, or impersonation attempts until it is too late. We wanted to build a real-time tool that can instantly warn users before they click a dangerous link or share sensitive information.

What it does

PhishGuard AI is a real-time phishing detection system that analyzes text messages, emails, and URLs to identify potential scams. Users can paste a message or link, and the system generates a risk score, threat level, and explanation. It highlights suspicious words, detects social engineering patterns, and analyzes URL structures to determine if content is likely malicious.

The system also includes:

Batch scanning for multiple messages
Threat analytics dashboard
Detection logs and reports
Chrome extension prototype simulation

How we built it

We built the system using Python and Streamlit for rapid prototyping and deployment. A machine learning classifier trained on phishing patterns works alongside a heuristic detection engine and URL risk analyzer. The final risk score combines these layers to provide an accurate and explainable result.

Challenges we ran into

The main challenge was building a multi-layer detection system within limited hackathon time while ensuring it remained stable and demo-ready. Balancing machine learning accuracy with real-time performance and usability was also a key challenge.

What we learned

We learned how to design a deployable cybersecurity tool under time constraints, integrate machine learning with rule-based systems, and build a clear, user-friendly interface for security applications.