AI Governance Check
Twelve questions on whether your AI rules are practised, not just written: owners, transparency, risk tiers, data, lifecycle and a stop button that works.
- Free
- No email, nothing sent
- 4 minutes
What this check measures
A policy document is not the same as governance. This check asks whether the rules are practised: does every system have a named owner, are use cases tiered by possible harm before they ship, and has anyone actually tried the stop path.
Twelve questions, two per pillar. Each answer maps to a level from 0 (ad hoc) to 4 (optimising). Your results and gap report appear on this page straight away. We ask for no email and send nothing anywhere.
AI Governance Check
Find your AI governance level
Two questions per pillar. Each answer maps to a level from 0 (ad hoc) to 4 (optimising). Results appear on this page and are not sent anywhere.
- AccountabilityA named owner for every model and agent in production
- TransparencyPeople can tell AI was involved and get a real explanation
- RiskUse cases tiered by possible harm before they ship
- DataSources, access and consent tracked like any regulated system
- LifecycleReviewed before launch, monitored, retrained or retired
- Human oversightA person can stop or override the system, and has tried
- 4 minutes
- 12 questions
- 6 pillars
- Nothing sent
How the level works
Three choices in the check are worth knowing before you read the result.
Answers are levels
Every answer is a level from 0 to 4. A pillar is the average of its two answers, and your level is the average of the six pillars, rounded half up.
Practice over paper
The top answers need evidence: tracked owners, reviewed tiers, a stop path that has been tested. A written policy on its own is level 1.
Frameworks as context
The gap report links your two weakest pillars to NIST AI RMF, ISO/IEC 42001 and the EU AI Act. It is a self-check, not an audit.
Questions
AI governance questions
How is my level calculated?
Each answer is a level from 0 (ad hoc) to 4 (optimising). A pillar is the average of its two answers, and your overall level is the average of the six pillars, rounded half up: an average of 2.5 counts as level 3. The overall score is that average as a share of 4.
Which frameworks does the report mention?
NIST AI RMF 1.0 (January 2023), ISO/IEC 42001:2023 (December 2023) and the EU AI Act, Regulation (EU) 2024/1689. As amended by the 2026 Digital Omnibus, its transparency duties apply from 2 August 2026, and high-risk duties from 2 December 2027 for stand-alone systems and 2 August 2028 for systems in regulated products.
Is this a compliance assessment?
No. It is a self-check of whether your rules are practised. The framework notes show where each pillar matters, but they do not tell you whether you comply with any law or standard.
Do I need to give my email?
No, and nothing is sent anywhere. The results and the gap report are made in your browser. “Email me this report” opens your own mail app with the report filled in, and you choose who gets it.
What should we do at level 0 or 1?
Start with an inventory: find every model, agent and bought AI tool in use, and name an owner for each. Rules written before you know what exists tend to stay on paper.
Can different teams score differently?
Yes, and that is worth knowing. Run the check once per team or business unit and compare the priority pillars.
Related tools
From gap report to working controls
We help you put owners, reviews and a tested stop path in place
Bring your gap report. We will go through the two priority pillars with you and say what it takes to close them.