1
A generated statistic is removed after the cited report cannot be found.
10 AI Fact-Check Checklist Examples
Use these examples to study structure, specificity, tone, and variation. They are demonstrations—not claims about a real person or organization unless the example itself makes that explicit.
Before you copy
What to notice in the examples
A strong fact-check process traces important claims back to authoritative sources, checks dates and scope, distinguishes inference from evidence, and never treats the model's confidence or citation formatting as proof.
- Extract factual claims from the draft.
- Classify each claim by verification risk and source requirement.
- Verify high-impact claims against primary or authoritative sources whenever available.
2
A product specification is checked against the current manufacturer documentation rather than a model summary.
3
A legal claim is narrowed after the cited rule applies only in one jurisdiction.
4
A scientific claim is rewritten because the paper supports an association, not causation.
5
A historical date is corrected after checking the archive catalogue.
6
A quotation is excluded because no primary or reliable source confirms the wording.
7
A current executive title is verified on the organization's official site before publication.
8
A comparison table is updated after discovering two figures came from different reporting periods.
9
A fact-check pass should treat a correct number with the wrong year as an error, even when the sentence sounds plausible and the source itself is real.
10
Mark every name, date, number, quotation, policy claim, and causal statement; verify each against a current authoritative source and record any qualification the draft omitted.
Keep the underlying decision or pattern, then replace the subject, evidence, relationship, constraints, and tone with details that belong to your situation. If your final line still works after swapping only one noun, it may be too close to the example.