We have co-produced a practical AI auditing catalogue and helper tool to support the assurance of AI in government.
Developed collaboratively with the audit offices of Brazil, Finland, Germany, Norway and the Netherlands, the catalogue acts as an auditors’ playbook for AI systems across their full lifecycle.
This catalogue is designed to support predictable, evidence‑based oversight. It will help public bodies:
- understand the key AI‑specific risks they need to manage
- build assurance into delivery, rather than bolting it on later
- avoid surprises at audit
What it covers
The catalogue sets out expectations across:
- governance and accountability
- data quality and management
- model development
- pre‑deployment evaluation
- change control
- live operation and production monitoring
The tool helps teams plan, test and evidence AI assurance without starting from scratch. It translates complex AI risks into clear, auditable expectations that fit with existing governance and delivery processes.
How to use the catalogue and tool
The catalogue identifies the AI‑specific risks public bodies need to manage and translates them into checklist‑based expectations at each stage of the lifecycle.
Alongside the catalogue, a helper tool simplifies evidence collection and sets out the controls auditors should expect across AI‑related risk areas.
Together, these practical tools are designed to support conversations about assurance, proportionate controls and the responsible use of AI in government.
