Skip to content
/ ScanIt Public

An open-source content authenticity framework merging multi-model AI detection (Fast-DetectGPT, GLTR) with immutable blockchain verification (Polygon/Ethereum) to solve the liability crisis in Generative AI.

License

Notifications You must be signed in to change notification settings

Didfu/ScanIt

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ScanIt: Content Authenticity Framework

License: Apache 2.0 Python 3.12+ Blockchain Status

ScanIt is an open-source framework designed to solve the liability crisis in the age of Generative AI. It merges multi-model AI detection with immutable blockchain ledgers to provide a transparent, mathematically verifiable chain of custody for digital documents.


📜 About This Repository

Note on Commit History: This repository contains the source code for the ScanIt framework as presented at the 2026 3rd International Conference on Integrated Intelligence and Communication Systems (ICIICS). The codebase was migrated from a restricted NVIDIA DGX Station A100 development environment; consequently, the granular development history is not preserved in this public release. This repository represents the "Camera-Ready" state of the system.


🚀 Key Features

1. Multi-Model Consensus Engine

Unlike "Black Box" detectors, ScanIt uses a transparent voting ensemble to reduce false positives:

  • Fast-DetectGPT: Provides rapid (<1.5s) initial filtering.
  • DetectGPT: Performs deep curvature analysis (Zero-Shot) for high-stakes verification.
  • GLTR: Analyzes token rankings to provide visual explainability.

2. Automated Source Tracer (Fact-Checking)

Detects AI hallucinations by verifying claims against the live web:

  • Uses a fine-tuned MT5 model to generate search queries from input text.
  • Cross-references claims using Google Fact Check Tools and Custom Search APIs.

3. Immutable Dispute Resolution

  • Non-Repudiation: Anchors the SHA-256 hash of every verification report onto the Polygon/Ethereum blockchain.
  • Decentralized Storage: Stores full analysis reports on IPFS, ensuring data persists even if the ScanIt servers go offline.

🛠️ Tech Stack

  • AI/Backend: Python 3.12, FastAPI, PyTorch, Transformers (Hugging Face)
  • Frontend: React.js, Tailwind CSS
  • Blockchain: Solidity (Smart Contracts), Truffle, Ganache
  • Infrastructure: IPFS (Kubo), Docker (Optional)

⚙️ Installation & Setup

Prerequisites: Python 3.12+, Node.js, and a GPU with at least 16GB VRAM (Recommended).

1. Clone the Repository

git clone [https://github.com/yourusername/ScanIt.git](https://github.com/yourusername/ScanIt.git)
cd ScanIt

2. Backend Setup

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

3. Blockchain Setup (Local Testnet)

Ensure you have Ganache installed and running on port 7545.

# Navigate to smart contracts folder
cd blockchain

# Compile and migrate contracts
truffle compile
truffle migrate --network development

Note: Copy the deployed contract address and update it in server/config.py.

4. IPFS Setup

# Start the IPFS daemon for decentralized storage
ipfs daemon

5. Run the Application

# Start the FastAPI Server
python server.py

The API will be available at http://localhost:8000.


⚖️ Acknowledgements & Attribution

ScanIt integrates and builds upon several open-source research projects. We strictly adhere to their licensing terms and attribute the original authors:

  • DetectGPT: Core curvature-based detection logic adapted from Mitchell et al. (2023). Original Repo
  • Fast-DetectGPT: Conditional probability scoring adapted from Bao et al. (2023). Original Repo
  • GLTR: Visual analysis overlays based on the GLTR framework by Gehrmann et al. (2019). Original Repo (Apache 2.0 License)

📖 Citation

If you use ScanIt in your research, please cite our ICIICS 2026 paper:

@inproceedings{scanit2026,
  title={ScanIt: An AI and Blockchain Powered Open-Source Framework for Content Authenticity},
  author={Mahyavanshi, Dhruv and Patil, Kartik and Bora, Gargi and Prajapati, Harsh and Agrawal, Shital and Kolekar, Uttam},
  booktitle={2026 3rd International Conference on Integrated Intelligence and Communication Systems (ICIICS)},
  year={2026},
  organization={IEEE}
}

📝 License

This project is licensed under the Apache-2.0 License - see the LICENSE file for details.

About

An open-source content authenticity framework merging multi-model AI detection (Fast-DetectGPT, GLTR) with immutable blockchain verification (Polygon/Ethereum) to solve the liability crisis in Generative AI.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages