Build a Verification Checklist for AI Answers and Their Citations
Paste any AI-generated answer and get a table of every claim and citation, rated by risk, with red flags and the exact check needed to verify each one.
You getClaim table with columns, top 5 checks to do first and claims too vague to verify as written
Prompt
You are a research librarian who audits AI-generated text before it goes into reports, articles or papers.
I will paste an AI-generated answer right after this message. Build a verification checklist for it. Do not confirm or deny claims from memory. Your job is to flag each claim and tell me how to check it.
Rules:
1. Extract every factual claim, number, date, name, quote and citation as its own row, quoted or closely paraphrased.
2. Rate each row High, Medium or Low risk. Any citation with authors, journal, DOI or page numbers, any direct quote, and any exact statistic or date is automatically High.
3. For each row, name the specific check: where to look (Google Scholar, Crossref DOI lookup, publisher site, official statistics office, original transcript) and what details must match.
4. Note red flags of fabrication: malformed DOIs, suspiciously round numbers, quotes with no venue or date, author and journal pairings that look unlikely.
5. Do not suggest replacement sources or rewrite the text.
Output format:
- Claim table with columns: #, Claim or citation, Type, Risk, Red flags, How to verify, Status (left blank)
- Top 5 checks to do first, in order
- Claims too vague to verify as written
End with one line on what this checklist cannot catch, such as a real source being quietly misrepresented.Example output
Tips and common edits
Why this works
- Telling the model not to verify from memory stops it from confidently 'confirming' its own hallucinations.
- Auto-rating citations, quotes and exact stats as High risk targets the exact places AI fabrication usually happens.
- A blank Status column and a ranked top-5 list turn the output into a working checklist, not just commentary.
Use before citing anything ChatGPT, Claude or Gemini wrote for you. Have the full AI answer ready to paste, including its reference list. Works best on answers with citations, stats or quotes.
Common edits
- Add a column for the APA or MLA citation as it should appear once verified.
- Ask it to only list High risk rows when the pasted answer is long.
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
Check a Website's Reliability with Lateral Reading Before You Cite It
Get a five-point lateral reading check of any site or organization, with search steps to verify each point and a clear cite, cite-with-caution or avoid verdict.
Check Whether a Health Headline Matches the Study Behind It
Paste a headline and its study abstract to get a 1-5 accuracy score, a check table on study design and risk figures, and a plain-English rewrite.
Check Whether a Statistic or Chart in an Article Is Used Honestly
Runs a six-point checklist on a statistic or chart you paste (base rate, denominator, timeframe, axis, range, causation) and rewrites the claim accurately.
Compare How Different Articles Report the Same News Event
Get a side-by-side table showing which facts all your pasted articles agree on, where they conflict, what each leaves out, and how each frames the story.
Fact-Check a Viral Post or Forwarded Message Claim by Claim Before Sharing
Splits a viral post into separate claims, rates each one, and gives you exact search phrases and source types to confirm or debunk it before you share.
Spot Logical Fallacies and Weak Arguments in Any Opinion Piece
Returns a ranked table that quotes each weak passage in an essay, names the fallacy, explains in one line why it fails, and suggests evidence that would fix it.
- Trace a Quote or Claim Back to Its Original Source Step by Step
- Break Down a News Article into Facts, Opinions and Loaded Wording
- Source Credibility Check with Claim-by-Claim Verification Plan
- Build a Literature Review Matrix from Pasted Abstracts with Gaps and Contradictions
- Code Interview Transcripts into Themes with Participant Counts and Verbatim Quotes
- Design a Bias-Checked Survey Where Every Question Informs a Decision