🚨 THE PROBLEM

Global digital fraud causes $1 Trillion in losses annually. In developing regions, non-technical users are constantly targeted by:

  • Fake Banking Receipts: Phishing scams that look 99% real but have subtle font errors.
  • Hidden Malware: Software installers (like KMS Activators) that bypass standard antivirus but contain visual malware artifacts.

💡 THE SOLUTION

VerifAI is a client-side defense system powered by Gemini 3.0 Computer Vision. Unlike traditional antivirus that scans code, VerifAI "sees" the fraud just like a human expert—only faster.

🎥 DEMO HIGHLIGHTS

  • Real-Time Detection: Instantly spots "Red Box" fraud (Phishing & Malware).
  • Multi-Platform: Works seamlessly on Desktop & Mobile (as shown in the video).
  • Resilient: Features a "Model Hunter" engine that ensures 100% uptime.

⚙️ HOW IT WORKS

  1. Upload: The user uploads a screenshot (Receipt, Email, or App Interface).
  2. Analyze: The app sends the image to gemini-3-flash-preview.
  3. Detect: The AI reasons through visual artifacts to classify as SAFE (Green) or SCAM (Red).

🛡️ SECURITY FOR JUDGES (JUDGE MODE)

To test this safely, we built a Secure Judge Mode:

  1. Open the App (via GitHub Pages or index.html).
  2. Enter your Google AI Studio API Key in the secure popup (No hardcoded keys!).
  3. Upload the test samples provided in the Repo to see Gemini 3.0 in action.

Built With

Share this project:

Updates