Procedure
AI Model Lifecycle Management Procedure
This procedure defines the end-to-end lifecycle management process for AI models, from initial development or procurement through deployment, monitoring, retraining, and decommissioning. It ensures that every AI model in the organization's inventory is governed, documented, and reviewed throughout its operational life.
- Type
- Procedure
- Version
- 1.0.12
- Date
- 2026-10-27
- Owner
- Chief Information Officer
- Approving authority
- Chief Information Officer
- Classification
- Public
- Confidentiality
- Public
- Category
- AI Governance
- Review frequency
- Annual and upon material AI framework or regulatory change
- Licence
- CC BY-SA 4.0
Read the full text on GitHub (opens in a new tab)
Outline
- Purpose
- Scope
- Roles and responsibilities
- 1. AI system inventory
- 2. Development and pre-deployment
- 3. Production deployment
- 4. Ongoing monitoring
- 5. Retraining and model updates
- 6. Annual review
- 7. Decommissioning
- Framework alignment
Frameworks it maps to
- ISO/IEC 42001:2023
- ISO/IEC 5338:2023
- OSFI Guideline E-23 (2025, effective 2027)
- EU AI Act (2024)
- NIST AI RMF 1.0 (2023)
- COBIT 2019
Related documents
- AI Governance Council Charterai
- AI Governance and Risk Frameworkai
- AI Testing, Validation and Documentation Standardai
- AI Audit Procedureai
- Acceptance Into Service Policysecurity
- Corrective and Preventive Action (CAPA) Procedurecompliance
- AI Model Succession and Identity Continuity Standardai
- Semantic Continuity in Migration and Retirement Standardgovernance
Next
All Documents