VerifAIed

A powerful desktop application for verifying online content and detecting misinformation using advanced AI technologies.

Features

1. Screenshot Analysis

• Instantly analyze any content on your screen
• Get real-time fact-checking and verification
• Receive detailed analysis with source citations
• Keyboard shortcut: ⌘+⇧+S

2. Recording Analysis

• Record a sequence of screen changes
• Automatically captures when content changes
• Analyzes the entire sequence for comprehensive verification
• Perfect for analyzing dynamic content or scrolling feeds
• Keyboard shortcut: ⌘+⇧+R

3. YouTube Analysis

• Analyze YouTube videos for misinformation
• Transcribes and fact-checks video content
• Provides comprehensive analysis of claims made
• Verifies statements against reliable sources
• Works with regular YouTube videos and Shorts

Technology Stack

• Frontend: PyQt5 for native desktop interface
• Image Processing: PIL for image optimization
• AI Analysis:
  • Together AI (Llama Vision) for image analysis
  • Perplexity API for fact verification
• Video Processing: youtube-transcript-api for transcription
• Output: Markdown to HTML rendering

Setup

1. Clone the repository
2. Install dependencies:

pip install -r requirements.txt

3. Set up environment variables in .env:

TOGETHER_API_KEY=your_together_api_key
PERPLEXITY_API_KEY=your_perplexity_api_key

Usage

1. Launch the application:

python main.py

2. Choose your analysis method:

   • Screenshot: Click "Take Screenshot" or press ⌘+⇧+S
   • Recording: Click "Start Recording" or press ⌘+⇧+R
   • YouTube: Click "Analyze YouTube" and enter video URL

3. View the analysis results with:

   • Main claims identified
   • Fact-checking results
   • Source citations
   • Overall credibility assessment

Requirements

• Python 3.8+
• macOS (primary development platform)
• API keys for Together AI and Perplexity
• Internet connection for AI analysis

Notes

• Screenshot and recording analysis use Together AI's Llama Vision model
• Fact verification performed by Perplexity's Llama-3.1-Sonar model
• YouTube analysis uses video transcripts for comprehensive verification
• All analyses include source citations for verification

Privacy & Security

• API keys stored securely in .env file
• No data stored permanently
• All analysis performed in real-time
• No user data collection

Future Development

• Enhanced error handling
• Additional video platform support
• Improved source verification
• Advanced AI model integration

For issues or suggestions, please open a GitHub issue.