Nova Guardian AI

Nova Guardian AI detects scam messages and phishing links using Amazon Nova. It analyzes text and screenshots to generate a risk score and help users identify online fraud.

Comment

Inspiration

Online scams and phishing attacks are becoming increasingly common. Many users receive suspicious messages, fake payment requests, or phishing links but cannot easily determine whether they are legitimate. I wanted to build a tool that helps users quickly analyze suspicious messages and screenshots to detect potential scams before they cause harm.

What it does

Nova Guardian AI is an AI-powered cybersecurity assistant that detects scam messages and phishing attempts. Users can paste suspicious text messages or upload screenshots of conversations or websites. The system analyzes the content and generates a risk score indicating whether the message is safe, suspicious, or potentially malicious.

The system can:

  • Analyze suspicious messages and URLs
  • Extract text from screenshots
  • Detect phishing patterns and scam language
  • Provide a risk score and explanation

How I built it

The project is built using Python and Flask for the backend and a modern web interface for user interaction.

Key components include:

  • Amazon Nova foundation models (via Amazon Bedrock) for reasoning and scam detection
  • OCR to extract text from screenshots
  • Python Flask API to handle requests
  • A web interface for entering messages and uploading images
  • AI-generated risk scoring and explanation system

The workflow is:

  1. User inputs text or uploads a screenshot.
  2. OCR extracts text from the image.
  3. The extracted text is sent to Amazon Nova.
  4. Nova analyzes the message for phishing patterns and suspicious intent.
  5. The system returns a risk score and explanation.

Challenges I ran into

One of the biggest challenges was integrating AI reasoning with real-world scam detection. Scam messages can vary widely in format and wording, so the system needs to analyze intent rather than just keywords. Designing prompts that allow the AI model to correctly identify phishing patterns was an important part of the process.

Another challenge was combining OCR with AI analysis so that screenshots could be processed effectively.

What I learned

Through this project I learned how to integrate Amazon Nova foundation models into an application, design prompts for AI reasoning, and build an end-to-end system that combines OCR, AI analysis, and a web interface.

What's next for Nova Guardian AI

Future improvements could include:

  • Browser extensions to detect scams in real time
  • Integration with messaging platforms
  • Automatic URL reputation checking
  • Improved phishing detection models
  • Mobile application support

Nova Guardian AI aims to help users stay safe online by identifying potential scams before they cause damage.

Built With

Share this project:

Updates