Audit drafts for confidently phrased factual hallucinations
Scans generated text for high-risk assertions, unverified statistics, and false certainty. Use it before publishing AI-written reports or summaries.
You getHigh-Risk Claims, neutralized Rewrites and verification Checklist
Prompt
Act as an investigative fact-checker. Audit the draft below to find claims that sound authoritative but carry a high risk of factual error or hallucination.
- Flag exact numbers, dates, statutory names, quotes, and comparative claims.
- Identify statements using definitive phrasing that should be framed conditionally.
- Highlight claims that require verification against original records before publication.
Output format:
- High-Risk Claims: a numbered list showing the verbatim claim, why it looks suspicious, and what to verify.
- Neutralized Rewrites: corrected phrasing replacing false certainty with measured language.
- Verification Checklist: 3 concrete primary sources to check.
Here is the draft:Example output
Paste your article or summary text at the end of the prompt in ChatGPT, Claude, or Gemini. Ask the model to suggest specific search queries for the two highest-risk claims found.
Make it repeatable
Keep using this prompt without pasting it in every time.
Prompt details
- By
- Prompts-GPT editors
- Published
Did this prompt work?
Keep exploring
More like this
Run a five point audit on an AI draft
Audits an AI-generated draft for hallucinations, tone inconsistencies, and unsupported claims before you hit send.
Split documented facts from AI interpretation and speculation
Separates verifiable source facts from analytical assumptions and extrapolations. Use it to keep executive memos, reports, and summaries strictly accountable.
Answer questions while highlighting unknown facts
Get accurate answers based strictly on your source text while preventing guesswork. Use it for contract review, technical specs, or detailed notes.
Audit AI text for hidden bias and one-sided framing
Detect subtle assumptions, suppressed counterarguments, and one-sided framing in an existing AI-generated answer.
Draft content with an internal self critique step
Produce higher quality drafts by forcing the model to critique its work against your constraints before printing the final version.
Ground every conclusion in exact source quotes
Extract verified facts backed by exact verbatim quotes from your text so you never mistake AI assumptions for source evidence.