transition2.aiStart a project

FREE TOOL · GOVERNANCE

AI risk-tier classifier

Answer eight questions about one AI system. You get a risk tier, the reasons for it, and the evidence that tier should need before release. Everything runs in your browser. Nothing you enter is sent or stored.

1. What does the system do with its output?
2. Can it write to records, send external messages, move money or change permissions without a person approving each action?
3. Does it make or support decisions about people in employment, education, credit, insurance, housing, health care, essential services, biometrics or migration?
4. Who sees its output or is affected by it?
5. What is the most sensitive data it can read?
6. Does it read untrusted content (web pages, inbound email, uploaded files) while it has tools it can call?
7. Is it a vendor AI feature switched on inside software you already use?
8. If it gets something wrong, can the effect be undone easily?

This is a planning aid, not legal advice. The tiers follow the model in Enterprise AI governance that unblocks delivery. Whether a law applies depends on your role, market and use; see the enterprise governance guides for dates and frameworks.

Related: all enterprise governance guides, AI acceptable use policy starter, and the Transition2 courses for teams putting this into practice.