AI Risk Review Prompt Template and Example

Reusable prompt

Identify AI risks and recommend controls before launch.

Task type: AI Risk Review
Objective: Identify AI risks and recommend controls before launch.

Context:
- [Project, product, or topic]: [Project, product, or topic]
- [Audience and situation]: [Audience and situation]
- [Constraints, must-haves, and things to avoid]: [Constraints, must-haves, and things to avoid]

Inputs to provide:
[Paste source material here]

Expected output:
1. Use case summary
2. Risk areas
3. Controls
4. Approval needs
5. Monitoring plan

Quality bar:
- Be specific and avoid generic advice.
- State assumptions explicitly.
- Prefer actionable next steps over broad theory.
- If important information is missing, ask up to 3 clarifying questions before answering.
- For time-sensitive or factual claims, label what is known, inferred, and needs verification.

Worked example

The example below fills the same prompt for a realistic Governance scenario. It is intentionally modest: the goal is to show how the prompt behaves, not to pretend one template solves every Governance problem.

Task type: AI Risk Review
Objective: Identify AI risks and recommend controls before launch.

Context:
- [Project, product, or topic]: A real Governance task using the AI Risk Review prompt
- [Audience and situation]: A teammate who needs a useful answer and clear next steps
- [Constraints, must-haves, and things to avoid]: Be specific, state assumptions, avoid unsupported claims, and keep the output easy to act on.

Inputs to provide:
Sample material: The team needs help with AI Risk Review. The current situation is messy, the goal is clear enough to start, and the answer should separate facts, assumptions, risks, and next actions.

Expected output:
1. Use case summary
2. Risk areas
3. Controls
4. Approval needs
5. Monitoring plan

Quality bar:
- Be specific and avoid generic advice.
- State assumptions explicitly.
- Prefer actionable next steps over broad theory.
- If important information is missing, ask up to 3 clarifying questions before answering.
- For time-sensitive or factual claims, label what is known, inferred, and needs verification.

How to use this prompt

1. Replace the placeholders with the actual AI Risk Review task, audience, source material, and constraints.
2. Keep the requested output sections unless you have a strong reason to remove one; they are there to make the AI answer easier to evaluate.
3. Paste the finished prompt into your AI assistant, then ask one follow-up question that tests assumptions or missing evidence.

What a good answer should contain

• 1. Use case summaryUse this section to make the answer concrete: Use case summary.
• 2. Risk areasUse this section to make the answer concrete: Risk areas.
• 3. ControlsUse this section to make the answer concrete: Controls.
• 4. Approval needsUse this section to make the answer concrete: Approval needs.
• 5. Monitoring planUse this section to make the answer concrete: Monitoring plan.

Why this prompt works

• AI Risk Review starts with an explicit task type and objective, which reduces vague answers.
• It asks for context, source material, and constraints before the model writes the final response.
• The 5 output sections make the answer scannable and easier to compare across attempts.
• The quality bar tells the assistant to ask clarifying questions and mark claims that need verification.

Common mistakes to avoid

• Leaving placeholders untouched and expecting the model to infer the missing context.
• Removing the output structure, then asking for a final answer that is hard to review.
• Using the prompt for time-sensitive facts without checking sources or dates.

Related prompt templates