Fake citations can slip into papers, literature reviews, and even published work—especially when references are copied from secondary sources or generated by AI. The good news: you can catch most “bad citations” in about 3 minutes.
Below is a practical checklist you can use before submitting a manuscript, sending a report to your team, or trusting a reference in AI-assisted writing.
What counts as a “fake citation”?
In practice, “fake citation” usually means one of these:
- Fabricated: the paper doesn’t exist.
- Mismatched: the paper exists, but the details don’t match (authors/year/journal/volume/pages).
- Misused: the paper exists, but it doesn’t support the claim you cited it for.
Think of it like verifying a product listing: a reference can look legitimate while the underlying identifier doesn’t match.
The 3-minute citation verification checklist
0:00–0:30 (30 seconds): Does the reference look normal?
Quick scan the reference for obvious issues:
- Does it have a plausible author list and year?
- Is the venue (journal/conference) a real, identifiable publication?
- Does it include an identifier like DOI / PMID / arXiv ID / ISBN (where applicable)?
Red flags
- Missing core fields (no journal, no year, no identifiers)
- Strange formatting (page range looks wrong, volume/issue missing for a journal article)
- Title sounds overly generic or marketing-like
0:30–1:30 (60 seconds): Can you find it in a trusted index?
Pick one trusted source and search by exact title (use quotes) or title + first author:
- PubMed (biomed/life sciences)
- Google Scholar (general)
- DOI lookup (publisher page or Crossref record)
Rule of thumb: if you can’t find the paper via title search, treat it as suspicious—especially for recent papers.
1:30–2:30 (60 seconds): Do the 4 key fields match?
Open the record you found and confirm:
- Title (key terms should match)
- First author
- Year
- Journal / conference
If two of these are clearly off, flag it for correction.
2:30–3:00 (30 seconds): Does it actually support the claim?
Most citation problems aren’t “the paper is fake”—they’re “the paper is irrelevant.”
Do a quick support check:
- Skim the abstract for the exact point you’re citing
- If you have the PDF, search the key term (Ctrl/Cmd + F)
- Watch for retractions or expressions of concern
If the evidence doesn’t support your wording (or direction of effect), don’t cite it.
Five high-probability patterns of bad citations
- Citation chaining: you cite A, but A only cites B (you never checked B)
- Correlation cited as causation
- Recent paper with no trace in major indexes
- Suspicious venue that’s hard to verify
- Reference-heavy paragraph where multiple citations are hard to find
Make this process effortless (optional)
If you validate citations frequently (systematic reviews, clinical writing, AI-assisted drafting), tools like LitSource can help you pre-screen citations/claims faster—so you spend time verifying the right things and keep every reference traceable.