Skip to content
JobPitch Studio
← All guides

AI governance and AI risk interview questions: how to answer with real examples

Krish Mootoosamy, founder of JobPitch Studio · 19 October 2026

If you work in risk, compliance, audit or privacy and you're moving into AI governance or AI risk work, your governance experience is probably your strongest asset. GRC practitioners we asked said the same thing: control frameworks, testing and policy writing carry over well. The harder parts are usually the technical questions and showing what you've actually done.

That second part matters in interviews. One recruiter told us they test CV claims by asking for a specific example and then a follow-up question. The claim gets you the conversation. The example is what they believe.

This guide covers nine questions worth preparing for, what each one is really testing, and what a strong example contains. It isn't a list of guaranteed questions, and every organisation will ask its own.

What a strong example contains

A useful structure is SOAR:

  • Situation: where you were and what was at stake, in a sentence or two.
  • Obstacle: what made it hard. This is where your judgement shows.
  • Action: what you did. Say “I”, not “we”, for your own part.
  • Result: what changed, with a number or an observable outcome where you have one.

Keep job titles accurate. If the work was AI governance but your title was “Compliance Manager”, say so. Practitioners told us that AI governance work often lands inside existing risk and compliance roles, so doing the work without the title is normal.

Nine questions to prepare for

1. Tell me about a time you assessed a risk in a system you didn't fully understand at first.

What it tests: whether you can get technical enough to describe what you’re assessing. One practitioner put it simply: you can’t assess a risk you can’t describe.

A strong example shows: what you didn't know, how you found out (who you asked, what you read, what you tested), and how that changed your assessment.

2. Tell me about a time you assessed a vendor or system that used AI or automated decisions.

What it tests: whether you look at the specific use, not “AI” as one thing.

A strong example shows: the practical questions you asked, such as what data went in, what the system could access, and whether it only suggested or could act on its own, and what you decided as a result.

3. Tell me about a time you put a control or policy in place before the rules were clear.

What it tests: governing under ambiguity. Some organisations are rolling out AI before they've decided how they'll use it.

A strong example shows: the interim step you chose, why it was proportionate, and how it changed as things became clearer.

4. Tell me about a time ownership of a risk was unclear across teams.

What it tests: working across privacy, security, model risk, legal and the business without authority over all of them.

A strong example shows: how you got to a decision on who owned what, and what happened to the risk in the meantime.

5. Tell me about a time you explained a technical risk to a non-technical audience.

What it tests: translation, for example explaining to legal or a board why an AI system producing a confident wrong answer is a different kind of error from a traditional model.

A strong example shows: who the audience was, what you simplified, and what decision they made afterwards.

6. Tell me about a time a business team wanted to move faster than the controls allowed.

What it tests: judgement and relationships. Can you say “not yet” in a way that keeps the work moving?

A strong example shows: the trade-off you proposed, not just the “no”.

7. Tell me about a time a control or assessment you designed was tested.

What it tests: whether your work holds up under audit, a regulator's review or an incident.

A strong example shows: what was tested, what held, what didn't, and what you changed.

8. Tell me about a time you made the case for resources or a new function.

What it tests: for senior roles especially, whether you can build a capability, not just run one.

A strong example shows: the argument you made, the evidence behind it, and the outcome.

9. Why AI governance, and what have you done to build your technical understanding?

What it tests: motivation and self-awareness.

A strong answer shows: concrete steps you've taken, honesty about what you're still learning, and how your existing experience connects.

A worked example

Illustrative

This example is synthetic. The person, organisation and numbers are made up to show the structure.

Question: Tell me about a time you assessed a vendor or system that used AI.

  • Situation: I was a compliance manager at a mid-sized general insurer. A claims software vendor added an AI feature in a routine update: it summarised claim notes and suggested a triage priority for each claim.
  • Obstacle: We had no AI policy yet. The vendor's documentation didn't say what data the feature used or whether the priority suggestion changed how claims were handled. Operations wanted to keep it on through a peak period.
  • Action: I proposed pausing the triage suggestion only and keeping the summaries, which carried less risk. I worked with IT and the vendor to answer three questions: what customer data went in and where it was processed, whether the feature could change claim records or only suggest, and who reviewed its output. I added a human check on summaries and wrote a one-page interim guardrail for AI features in existing vendor tools, which the risk committee approved.
  • Result: Summaries stayed live with a human check. The triage suggestion came back six weeks later after a sample review of 200 claims. The interim guardrail became the starting point for our AI policy, and the vendor contract now requires notice before new AI features are switched on.

Why it works: the action is clearly the speaker's own, the questions asked are specific, there's a trade-off rather than a blanket “no”, and the result is observable.

A weaker version of the same story: “I was responsible for AI governance across our vendor portfolio.” It’s a claim with nothing an interviewer can test.

If you don’t have an AI example yet

That’s common, and it isn’t a reason to stretch the truth. Look for adjacent work: a new technology you assessed, automated decision rules, model risk, outsourcing or a vendor change. Say plainly that the system wasn’t AI, then show the same skills. Pair it with a clear answer to question 9 about what you’re doing to close the gap.

Missing a question you’re preparing for?

Send it to support@jobpitch.studio. We’ll add the most useful ones to this guide, without naming you. Please leave out your employer’s name and anything else that could identify you.

Build your examples in JobPitch Studio

JobPitch Studio helps you build a story bank of your achievements, one field at a time, from facts you confirm. When you’re preparing, paste an interview question into Question Match to see which of your stories fits best and why it was chosen.

Try it on a real role

Free while in beta