FairTrain

FairTrain is an open protocol that fundamentally changes how AI models ingest training data by turning creators' work into a royalty-generating asset. Today, over 50 million professional creators (artists, writers, photographers, musicians) have their content scraped by AI companies with zero consent and zero compensation. Meanwhile, the EU AI Act mandates strict transparency, and AI labs face massive lawsuits. FairTrain solves this by providing a universal, censorship-resistant namespace where creators can publish machine-readable licensing terms and prices per token. When an AI crawler downloads a creator's work, it automatically discovers these terms via ENS, verifies the content's provenance on 0G Storage, and triggers an autonomous micro-payment streamed directly to the creator's wallet. Every transaction is logged immutably, generating a complete compliance report for regulators instantly.

FairTrain is built on a composable Web3 stack integrating ENS, 0G, and KeeperHub to create a seamless, load-bearing infrastructure.

ENS: Creator identities and licensing terms (like price per 1000 tokens and permitted uses) are strictly formatted and stored in ENS text records. This acts as a censorship-resistant discovery API queryable via viem. 0G Storage: Content is hashed locally and linked to 0G Storage and KV. This ensures a tamper-proof, timestamped audit trail that courtrooms and regulators can trust for AI Act compliance. KeeperHub & Smart Contracts: Our FairTrainRegistry.sol and RoyaltyStream.sol contracts run on the 0G testnet. When an AI pipeline ingests data, the contract emits a DataConsumed event. KeeperHub acts as a milestone-triggered escrow engine; it listens for these events and guarantees high-frequency USDC micro-payments.

A React/Vite frontend serves as the creator dashboard to mint and track these assets. For the AI labs, we provide @fairtrain/pipeline, a lightweight Node SDK allowing them to become entirely compliant by adding literally one line of code to their data-fetching scripts.