Practical guide
AI use register template
Record actual AI use in a practical way without presenting the record as a compliance certificate.
What the register should answer
A useful register shows where AI is used, why it is used, who is responsible, which data categories are involved and how the result is checked. The goal is operational visibility, not a one-time inventory that nobody updates.
- Use case and purpose
- Tool or model
- Owner and approver
- Data categories
- Human check
- Open actions and review date
Keep evidence separate from assumptions
Record what actually happens in the workflow. Mark examples as examples and open questions as open. Do not label an entry compliant, safe or low risk unless a qualified assessment supports that conclusion.
Review the register regularly
A short recurring review is more useful than a large initial exercise. Check whether the tool, data, purpose, owner or approval step changed and whether an open action can be closed.