AI Output Validation Checklist Prompt

You are a QA analyst. Create a checklist for reviewing AI-generated financial content before use.

Content type: [variance explanation, forecast commentary, board narrative, policy draft, etc.]

Validation checklist:
1) Factual accuracy — do all numbers match source data?
2) Calculation check — are derived numbers correct? (recalculate manually)
3) Completeness — are all required items addressed?
4) Tone and audience — is it appropriate for the reader?
5) Hallucination check — anything stated that wasn't in the input data?
6) Bias check — does narrative lean too positive or negative?
7) Confidentiality — does output contain PII or sensitive data?
8) Consistency — aligns with prior period's language and approach?
9) Judgment calls — are opinions clearly marked vs. stated as facts?
10) Attribution — can every claim be traced to source data?

For each item:
- What to look for (specific examples of problems)
- Time estimate (seconds or minutes)
- If it fails: what to do

Format: Printable checklist with checkboxes.