ChatGPT Upload Guard

A Chrome extension that detects prompt injection attacks in documents uploaded to ChatGPT using local AI analysis.

Install from Chrome Web Store

Demo

demo-upload-guard.mov

Features

• Real-time Scanning - Automatically scans document uploads to ChatGPT
• Local AI Analysis - Private, in-browser threat detection using WebLLM
• Smart Detection - Identifies prompt injections, jailbreaks, social engineering, and data extraction attempts
• Privacy-focused - All processing happens locally in your browser

Installation

From Chrome Web Store (Recommended)

Install ChatGPT Upload Guard

From Source

1. Clone and build:

   git clone https://github.com/netanelavr/chatgpt-upload-guard.git
   cd chatgpt-upload-guard
   npm install
   npm run build

2. Load in Chrome:

   • Go to chrome://extensions/
   • Enable "Developer mode"
   • Click "Load unpacked" and select the project folder

Usage

1. Navigate to ChatGPT
2. Upload a document (.txt, .docx, .md, etc.)
3. The extension automatically scans for threats
4. Review results and choose to block or proceed with upload

Supported Files

• Documents: .txt, .docx, .pdf, .md, .csv
• Code: .js, .ts, .py, .java, .cpp, .html, .css, and 30+ more
• Config: .json, .yaml, .xml, .env

Privacy & Security

• 100% Local: All analysis happens in your browser
• No Data Sent: Files never leave your device
• Open Source: Full code available for audit

Development

npm install      # Install dependencies
npm run dev      # Development build with watch
npm run build    # Production build

Limitations

• Requires WebGPU support
• Initial model download (~1-2GB) on first use
• Only works on ChatGPT websites

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see LICENSE for details.

---

Stay safe online! Always be cautious when uploading sensitive documents to AI services.