Veru

AI Citation Auditor & Hallucination Detector

Verify ChatGPT/Claude citations against real academic databases instantly.

Visit Live Demo on veru.app »

📖 About The Project

LLMs like ChatGPT often hallucinate academic citations—inventing papers that sound real but don't exist. Veru acts as a forensic auditor for your research.

It extracts citations from your text and cross-references them against massive academic databases to verify their existence and content accuracy.

Demo

Key Features

• 🕵️‍♂️ Anti-Hallucination Extraction: Uses Gemini 2.0 to strictly extract citations without "autocorrecting" user errors.
• 📚 Multi-Database Verification:
  • Tier 1: Checks OpenAlex (250M+ works) with smart title matching.
  • Tier 2: Fallback to Semantic Scholar if OpenAlex misses.
  • Tier 3: Final forensic sweep using Google Search Grounding.
• 🧠 Content Consistency Check: Compares the user's claim against the actual abstract to detect mismatched summaries ("stitched" hallucinations).
• 💾 Local History: Automatically saves your audit sessions locally (no login required).

🧠 How It Works (The Audit Pipeline)

graph TD
    A[User Input] --> B(LLM Extraction)
    B --> C{OpenAlex Search}
    C -->|Found & High Confidence| D[✅ Verfied Real]
    C -->|Not Found| E{Semantic Scholar}
    E -->|Found| D
    E -->|Not Found| F{Google Search Grounding}
    F -->|Verified| D
    F -->|Hallucinated| G[🛑 MARK AS FAKE]
    D --> H[Content Consistency Check]
    H -->|Match| I[Final Output]
    H -->|Mismatch| J[⚠️ Content Warning]

Loading

🛠 Tech Stack

• Frontend: Next.js 14 (App Router), Tailwind CSS, Lucide React
• Backend: Python FastAPI, Uvicorn, SlowAPI (Rate Limiting)
• AI & Data: Google Gemini 2.0 Flash, OpenAlex API, Semantic Scholar API
• Deployment: Vercel (Frontend) + Render (Backend)

🚀 Getting Started

Follow these steps to run Veru locally.

Prerequisites

• Node.js 18+
• Python 3.9+
• A Google Gemini API Key (Get one from Google AI Studio)

Installation

1. Clone the repo

   git clone https://github.com/Yinghao-Guan/Veru.git
   cd Veru

2. Setup Backend

   cd audit_backend
   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`
   pip install -r requirements.txt
   
   # Create .env file
   echo "GEMINI_API_KEY=your_api_key_here" > .env
   
   # Run server
   python main.py

   The backend will run on http://localhost:8000.

3. Setup Frontend Open a new terminal:

   cd frontend
   npm install
   
   # Run development server
   npm run dev

   The frontend will run on http://localhost:3000.

🛡️ Security

• Rate Limiting: Backend is protected by slowapi to prevent abuse.
• CORS: Restricted to production domains and localhost.
• No Data Retention: We do not store user queries on our servers (history is local-only).

🤝 Contributing

Contributions are what make the open source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated.

1. Fork the Project
2. Create your Feature Branch (git checkout -b feature/AmazingFeature)
3. Commit your Changes (git commit -m 'Add some AmazingFeature')
4. Push to the Branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

📄 License

Distributed under the MIT License. See LICENSE for more information.

🙏 Acknowledgments

• OpenAlex for their incredible open data.
• Semantic Scholar for the academic graph API.
• Google Gemini for the reasoning engine.

If you find this project interesting, please give it a ⭐️ on GitHub! It helps a lot.