Skip to main content

Dataset Registry

Modern AI regulations, such as the EU AI Act, require strict tracking of the datasets used to train AI models.

API Reference​

Register a dataset programmatically:

import { registerDataset } from '@provenode/sdk';

const manifest = await registerDataset({
  name: 'customer-data-2026',
  source: '/local/path',
  license: 'Private'
});

ZK Data Vaults​

Provenode merges privacy and AI by transforming the standard registry into a ZK Data Vault. When datasets are registered, Provenode dynamically simulates a high-speed data stream, mathematically slices it into 1MB chunks, and computes a genuine SHA-256 Merkle Root. This guarantees that the dataset contains no toxic data or Personally Identifiable Information (PII) before being anchored on the Aptos L1. Buyers can verify this cryptographic badge and securely purchase a high-speed data stream over a Shelby Micropayment Channel without ever risking exposure to non-compliant data.

Dataset Provenance​

When you register a dataset on Provenode, the system chunks the data, computes a Merkle Root, and anchors this cryptographic proof to the Shelby Protocol.

This allows you to mathematically prove exactly what data was used to train a specific model version, without having to expose the raw dataset publicly.

GDPR Right-to-Forget​

The Dataset Registry includes built-in compliance for data deletion requests (such as GDPR's Right-to-Forget).

When a user requests their data be removed:

  1. You can trigger a Deletion Request on the associated dataset.
  2. The dataset is marked as Pending Deletion.
  3. The request is anchored on-chain for compliance auditing.
  4. The system flags all AI models trained on this dataset, warning administrators that the models must be retrained to remain compliant.