Last updated:
A documented record of the origin, ownership, and lifecycle of a dataset used to train or fine-tune an AI model.
Provenance provides the audit trail for data. It tracks where the data was sourced (e.g., public web, licensed archive, user-generated), what licenses were attached, and how it was transformed. High-quality data provenance is essential for demonstrating legitimate interest under GDPR and defending against copyright claims.
Real world example:
An AI startup maintains a detailed ledger showing that all 10 million medical images used to train their diagnostic tool were de-identified and used with explicit institutional consent - satisfying a buyer's due diligence requirements.
