Procedure
Training Data Governance Procedure
This procedure governs the lifecycle of data used to train, fine-tune, or otherwise shape model behaviour. It addresses sourcing, lawful basis, consent, sensitive-content removal, lineage, deletion propagation, supplier disclosure, and the recurring obligations that arise when a model derived from a dataset remains in use after the dataset itself has changed.
- Type
- Procedure
- Version
- 0.0.16
- Date
- 2026-10-30
- Owner
- AI Data Steward
- Approving authority
- Governance Library Maintainer
- Classification
- Public
- Confidentiality
- Public
- Category
- AI Governance
- Review frequency
- Annual and upon material change to training-data sources, regulation, or supplier landscape
- Licence
- CC BY-SA 4.0
Read the full text on GitHub (opens in a new tab)
Outline
- Purpose
- Scope
- Procedure
- Coordination with adjacent governance
- Framework alignment
- Limitations
Frameworks it maps to
- EU AI Act
- EU GPAI Code of Practice (Copyright Chapter, July 2025)
- ISO/IEC 42001:2023
- GDPR (EU) / UK GDPR
- LGPD
- PIPL
- NIST AI RMF 1.0
- ISO/IEC 5259-2:2024
- ISO/IEC 5259-3:2024
- ISO/IEC 5259-4:2024
- ISO/IEC 5259-5:2025
- ISO/IEC 8183:2023
- OECD AI Principles
Related documents
- AI Governance and Risk Frameworkai
- AI Security and Risk Standardai
- AI Model Risk Standardai
- Dataset Datasheet Templateai
- Model Registryai
- AI System Impact Assessment Procedureai
- Privacy and Data Governance Policyprivacy
- Pseudonymization and Anonymization Standardprivacy
- Data Subject Rights Management Procedureprivacy
Next
All Documents