SIGN.AI

SIGN.AI

A blockchain-based AI content verification system that provides cryptographic proof of AI-generated content authenticity.

Project description

SIGN.AI is a decentralized system that allows anyone to verify if content was truly generated by an AI model—and detect if it has been edited or tampered with.

It works by capturing the entire content generation event (prompt, output, model parameters), calculating cryptographic hashes, and packaging them into a signed provenance object. This object is then signed using EIP-712, stored on IPFS, and can later be verified through zero-knowledge proofs (ZKPs) to prove correctness without revealing sensitive data.

It combines Next.js frontend, Ethereum wallet signing, RISC Zero ZK circuits, and IPFS content-addressable storage to make AI content verifiable, tamper-evident, and transparent. This helps combat misinformation, deepfakes, and plagiarism while preserving privacy.

Technologies used

• Next.js 14 + React (frontend)

• RainbowKit + Wagmi (Ethereum wallet integration)

• TypeScript

• RISC Zero (ZK proof generation)

• IPFS (content storage)

• SHA-256 (content hashing)

• EIP-712 (structured data signing)

• Docker (deployment)

Basic architecture

• Frontend: Next.js SPA with wallet integration and real-time hash previews

• Backend APIs: Deterministic keyword generation engine (mock AI), provenance signing, and verification endpoints

• ZK Proof Layer: RISC Zero hosted prover generating keyword proof

• Storage Layer: Mock IPFS-based content-addressable storage used for local development.

• Verification Layer: UI panels to verify signed provenance by CID

Source code

• GitHub Repository

Deployment

• Live Demo(ZK Proofs only available on local)