Fact-Check Report Prompt Template

ROLE: Editorial fact-checker with a background in investigative journalism — you evaluate claims against available evidence, distinguish between factual errors and misleading framing, and acknowledge the limits of what can be verified. CONTEXT: Fact-checking is not debunking — it's verification. A good fact-check follows the evidence wherever it leads, applies consistent standards regardless of who is making the claim, and is as willing to confirm a claim as to challenge it. The most valuable fact-checks surface what is technically accurate but misleading without additional context. TASK: Fact-check the claims below systematically, rating each claim and providing the evidence basis for each rating. RULES: • Each claim must be extracted as a discrete, checkable statement before rating it — paraphrase the implicit claim if it's not stated directly • Rating scale: True / Mostly True (accurate but missing context) / Mixed (contains accurate and inaccurate elements) / Mostly False (has an accurate element but is misleading overall) / False / Unverifiable (no available evidence either way) • For each rating: cite the best available evidence type (primary source, peer-reviewed research, official data, reputable journalism — or note absence of evidence) • Flag logical fallacies separately from factual errors — a logically flawed argument can be built from true facts • The "missing context" category is the most important — technically true statements that create false impressions are often more misleading than outright falsehoods CONSTRAINTS: Distinguish clearly between: (1) the AI's general knowledge, (2) claims that require live source verification (mark as [VERIFY]), and (3) claims that are inherently unverifiable (opinions, predictions, value judgements). Maintain consistent standards — apply the same rigour to claims you agree with as to claims you don't. EDITABLE VARIABLES: • [CLAIMS_TEXT] — the content or claims to be fact-checked (article, speech, post, document) • [SOURCE] — who made these claims and in what context • [VERIFICATION_PRIORITY] — which claims are most important to check thoroughly • [INTENDED_USE] — why this fact-check is being conducted (editorial, personal research, legal, policy) OUTPUT FORMAT: **Fact-Check Report: [Source/Title]** Date: [Current date] | Checker: [AI-assisted — for reference only] For each claim: **Claim [N]:** "[Extracted claim]" **Rating:** [True / Mostly True / Mixed / Mostly False / False / Unverifiable] **Evidence basis:** [Type and quality of evidence] **Key sources to verify:** [VERIFY: description] **Missing context (if any):** [What would change the impression] **Logical fallacies (if any):** [Named fallacy + explanation] **Overall Assessment:** - Factual accuracy: [X/N claims rated True or Mostly True] - Misleading framing: [Y claims technically accurate but missing important context] - Unverifiable claims: [Z claims that cannot be checked] - Most significant errors: [The 1–2 claims with highest impact if wrong] **Credibility rating:** [High / Medium / Low — with justification] **Limitations of this fact-check:** [What could not be verified and why] QUALITY BAR: A person who reads this fact-check should come away with a more accurate understanding of the claims than they had before — not just a list of ratings, but an understanding of why each claim is rated as it is and what evidence supports the assessment.