An AI-powered fact-checking application that automatically transcribes, analyzes, and verifies claims from videos, Instagram content, and text. This tool helps combat misinformation by providing evidence-based verification with accurate source references.
Repository: https://github.com/owgit/factCheck
- Multi-Source Verification: Process videos, Instagram posts, and free text input
- AI Transcription: Convert speech to text using OpenAI's Whisper model
- Advanced Fact Checking: Verify claims using OpenAI's GPT-4o with web search capabilities
- Claim Detection: Automatically extract and analyze factual statements
- Image Analysis: Extract and analyze visual content from videos and images
- Detailed Reports: Get comprehensive fact-check reports with sources and reliability scores
- Multilingual Support: Automatically detects input language and responds accordingly
- Web Search Integration: Enhances verification accuracy with real-time web information
- User API Keys: Users can provide their own OpenAI API key to use their own credits
- Monetization Ready: Built-in ad placeholders with skyscraper ads on both sides
- Responsive Design: Optimized for all screen sizes with a wider content area on large screens
- Python with FastAPI
- OpenAI API integration (GPT-4o, Whisper)
- MoviePy for video processing
- Instagram content extraction
- React with user-friendly tabbed interface
- Tailwind CSS for styling
- Responsive design with skyscraper ad support
- User-provided API key management
- Docker and Docker Compose for containerization
- Environment-based configuration
- Python 3.9+
- Node.js and npm
- OpenAI API key
- FFmpeg (for video processing)
- Docker and Docker Compose (optional)
-
Clone the repository
git clone https://github.com/owgit/factCheck.git cd factCheck -
Copy
env.exampleto.envand add your API keyscp env.example .env
-
Choose your setup method:
Docker Setup (Recommended)
docker-compose up -d
Access at http://localhost
Manual Setup
# Install backend dependencies pip install -r requirements.txt # Start backend cd video-upload-app uvicorn app:app --reload # In another terminal, install and start frontend cd video-transcription-frontend npm install npm start
Access at http://localhost:3000
Key environment variables:
| Variable | Description | Default |
|---|---|---|
OPENAI_API_KEY |
Your OpenAI API key (users can also input their own in the UI) | required for server |
FACT_CHECK_MODEL |
OpenAI model for fact checking | chatgpt-4o-latest |
IMAGE_ANALYSIS_MODEL |
OpenAI model for image analysis | gpt-4o-mini |
TRANSCRIPTION_MODEL |
OpenAI model for transcription | whisper-1 |
WEB_SEARCH_MODEL |
Model used for web searches | gpt-4o-search-preview |
USE_WEB_SEARCH |
Enable web search during fact checking | true |
INSTAGRAM_USERNAME |
Instagram username for content extraction | optional |
INSTAGRAM_PASSWORD |
Instagram password for content extraction | optional |
See .env.example for all configuration options.
The application includes built-in ad placeholders for easy monetization:
- Skyscraper Ads: 160px × 600px ads on both sides of the content
- Leaderboard Ads: 728px × 90px ads at the top and bottom
- In-Content Ads: Banner ads (468px × 60px) between content sections
These placeholders can be easily replaced with actual ad code from ad networks like Google AdSense.
Users can now provide their own OpenAI API key in the UI:
- Secure input with show/hide functionality
- Local storage to remember the key between sessions
- Validation to ensure proper key format (must start with "sk-")
- Visual feedback for saved and invalid states
- All API requests include the user's key when provided
This feature allows users to use their own OpenAI credits instead of the server's.
- Input Processing: Upload a video, provide an Instagram URL, or enter text
- Text Extraction: For videos, speech is transcribed to text using Whisper
- Fact Identification: The system identifies key factual claims in the content
- Verification: Each claim is verified against reliable sources and web information
- Report Generation: A comprehensive fact-check report is generated with sources and reliability ratings
Instagram integration may encounter issues due to Instagram's measures against automated access:
- 401 Unauthorized: Try different credentials or manual upload
- No media found: Content might be private or inaccessible
- Page structure errors: Instagram may have changed their page structure
If you encounter Docker-related problems:
# Debug Docker setup
./docker-debug.sh
# Check logs
docker-compose logs -f
# Rebuild containers
docker-compose up -d --buildIf the web search functionality isn't working:
- Verify your OpenAI API key has access to search-enabled models
- Check that
WEB_SEARCH_MODELis set to a search-capable model (e.g.,gpt-4o-search-preview) - Ensure
USE_WEB_SEARCHis set totruein your environment
If the user API key functionality isn't working:
- Make sure your backend server is configured to accept the
X-OpenAI-Api-Keyheader - Verify that the OpenAI key format is correct (should start with "sk-")
- Try clearing browser storage and re-entering the key
Contributions are welcome! Please feel free to submit a Pull Request.
- Fork the repository
- Create your feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add some amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
This project is licensed under the MIT License, which allows you to use, modify, and distribute the code freely.
While not required by the license, if you use this repository, the author would appreciate:
- Including a credit to Uygar Duzgun in your project
- Adding a link back to https://uygarduzgun.com
- Letting the author know how you're using the project
- Uygar Duzgun - uygarduzgun.com
- GitHub: owgit
- Support: Buy me a coffee