Free Template · Documentation
Model Documentation Template
Auditors and risk teams will ask the same questions: what is the model for, what data was used, how was it tested, and what controls are in place? Use this template to document models consistently and reduce governance friction.
What to document
Model overview
- •Model name and version
- •Owner / accountable team
- •Intended use and target users
- •Out-of-scope / prohibited use
- •Deployment context (internal, customer-facing, embedded)
Training data and provenance
- •Data sources and collection method
- •Data licensing and usage rights
- •PII / sensitive data handling
- •Data representativeness and known gaps
- •Data retention and deletion policy
Evaluation
- •Metrics used (accuracy, F1, calibration, hallucination rate)
- •Test datasets and methodology
- •Bias/fairness testing results
- •Security testing (prompt injection, adversarial inputs)
- •Human review process and acceptance criteria
Limitations and risks
- •Known failure modes
- •High-risk scenarios
- •Residual risk acceptance and approver
- •User communication requirements (disclosures)
Controls, monitoring, and change management
- •Guardrails and policy controls
- •Logging and audit trail location
- •Monitoring metrics and alert thresholds
- •Review cadence (30 days / quarterly / annually)
- •Model update process and rollback plan
Complete AI governance library
Get all 39 templates in one ZIP
Policies, registers, checklists and rollout plans, ready to customise.
- 46 files (DOCX, XLSX, PPTX)
- 2.0 MB total
- Updated October 2026
- NIST AI RMF · ISO 42001 · EU AI Act aligned
Work email only. Aona AI provides a SOC 2 Type 2 report for security review.
Get started
Need governance evidence automatically?
A template helps you document point-in-time. Aona AI gives you continuous discovery, policy enforcement, and audit-ready evidence.