Learning path · Enterprise Patterns & Governance · 89
AI Governance
Policies, roles, and review boards governing model selection, data use, eval evidence, and incident response.
Why it matters
- Required for regulated industries and enterprise procurement.
- Clarifies who approves new tools and datasets.
- Connects red teaming and evals to release gates.
Key ideas
- Risk tiers
- Approval workflows
- Model inventory
AI governance maintains model inventories, risk classifications, and documentation for auditors—intended use, eval results, known failures. High-risk features pass legal and security review with rollback plans. Incidents trigger root cause across data, prompts, and tools—not blame on a single engineer. Maintain a living risk register linking models, datasets, and incidents so audit questions do not require archaeology across Slack. Validate changes on production-like eval slices before rollout. Link governance tickets to model versions and dataset hashes for reproducible audits. Link governance tickets to model versions and dataset hashes for reproducible audits.
Updated 2026-08-09 · Full learning path