download dots
AI PromptsCategoriesAI Governance

⚖️ Assess the Risk Before You Approve

Put a sensible process around how your organization uses AI, without stopping anyone getting work done. These prompts build the approval queue, the register of what is running, the log of what each tool actually did, and the policy an assistant checks submissions against. Written for the person who has been asked to own this and given no software to do it with.
  • Tested prompts
  • Edit for your use case
  • Works across models
  • Save in Taskade
Assess the Risk Before You Approve

Copy

We approve AI use cases inconsistently depending on who reviews them. Build me a risk assessment app. Keep a standard question set as editable records covering the data involved, whether it includes personal or confidential information, whether a decision affects people, whether a human reviews the output, where the model runs, what happens if the output is wrong, and how it would be detected. Score impact and likelihood separately and combine them into a risk level with a plain explanation of what that level means and what it requires. Record the controls that reduce each risk, and attach them as conditions of any approval so an approval is never unconditional. Add an AI agent that drafts the assessment from the submission, names the risk that actually matters rather than listing every theoretical one, and flags where a proposed control would not really address the risk it is attached to. Automate two things: a reassessment reminder when a use case changes materially, and a scheduled annual review of all high risk cases. Two logins: a submitter who completes their own assessment, and a reviewer who scores and approves.

Run this prompt with AI
STATUS · LIVE
v1

01 / Spark - intelligence awakens

Assess the Risk Before You Approve

Run a standard risk assessment on every AI proposal scoring impact and likelihood, with controls recorded against approval.

Start with AI
// the spark catches · intelligence awakens · the answer unfolds

Risk judged case by case is judged inconsistently. A standard assessment makes decisions comparable, and comparable decisions are the ones you can defend.

What's Included

  • A standard question set applied to every proposal.
  • Risk from impact and likelihood rather than a single impression.
  • Controls recorded against risks so approval has conditions.
  • An assistant that drafts the assessment and names the real risk.

How To Use

  1. Use the same questions every time. Consistency is what makes it defensible.
  2. Assess impact and likelihood separately, then combine them.
  3. Attach controls to approvals. An approval without conditions is a gamble.
  4. Reassess when the use case changes, not only at first approval.

Submit through the governance office, and check it against the policy register.

Discover more

More Governance and Oversight Prompts

View all Governance and Oversight Prompts