SAAF Ethical Risk Pre-Assessment

Time to Complete: 15–25 minutes per use case

Purpose: Identify and flag early risks related to bias, privacy, misuse, and harm before building or deploying AI capabilities.

Why This Matters

Even well-meaning AI projects can lead to unintended harm. This assessment helps you pause before launch, ask the right questions, and put safeguards in place. It ensures AI, custom agents, or any AI solution aligns with your organization’s values and legal obligations.

Section 1: Use Case Overview

Use Case Title:
Purpose of the AI Use:
Tools Involved:
Who Will Use This AI Capability?
Who Will Be Affected By It?

Section 2: Risk Indicators

Check each item that may apply to your use case.

Privacy & Data Use

Question	Yes / No / Unsure
Will this AI process personal, sensitive, or regulated data (e.g., PII)?
Will the data come from user-generated or public-facing content?
Is there transparency about what data is being used?

Fairness & Bias

Question	Yes / No / Unsure
Could the AI reinforce stereotypes or exclude certain groups?
Is the training data diverse and representative of the people affected?
Would different users receive different outcomes for similar inputs?

Autonomy & Human Oversight

Question	Yes / No / Unsure
Will humans review the AI’s output before action is taken?
Could users be misled into thinking the AI’s answer is final or neutral?
Is it clear when content is AI-generated?

Harm & Reputational Risk

Question	Yes / No / Unsure
Could incorrect output cause harm (e.g., misinformation, legal risk)?
Could this be used outside its intended purpose or audience?
Would it be damaging if this output appeared in public or in the news?

Section 3: Risk Summary Score (Optional)

Assign a score for each area:

1 = Low Concern 

2 = Some Concern 

3 = High Risk / Unclear

Risk Domain	Score (1–3)
Privacy & Data Use
Fairness & Bias
Autonomy & Oversight
Reputational Harm

Total Risk Score: ___ /

124–6 = Low Risk → Proceed with documentation

7–9 = Medium Risk → Add safeguards before moving forward

10–12 = High Risk → Escalate to governance board or legal

Section 4: Risk Mitigation Plan

For any “Yes” or “High Risk” item, describe your mitigation plan:

Risk or Concern Identified	How You’ll Address It (mitigation, policy, review)

Section 5: Required Follow-Ups

☐ Add human-in-the-loop review step

☐ Add user-facing disclaimers or AI attribution

☐ Involve legal/compliance for review

☐ Postpone or redesign use case☐ Other: __________________________________________

Sign-Off & Next Steps

Reviewed By	Role	Date

Store this form in your AI Governance documentation or link it to the related Jira Epic or Confluence page.