Hallucinated citations: what they are and how to spot them
By Hallucite · Content reviewed
Check references against publication records, whether you are reviewing someone else’s work or your own. Treat “hallucinated citations” as a publication-verification question: can you establish the work described by the reference, and do its details match? Research such as GhostCite uses the term for unreliable or fabricated references. Your task is to investigate the evidence for an entry, without guessing who wrote it or why.
Define the problem before using the label
Keep different outcomes in separate notes. A reference may need a spelling correction, may point to a different publication through its identifier, or may remain unresolved after searching. Investigate which situation you have before calling an entry fabricated. The citation-verification evaluation discusses the limits of detecting problematic references; it is a reason to inspect findings carefully rather than turn a search failure into a final accusation.
For your own bibliography, a useful question is “which publication did I mean to cite?” For someone else’s bibliography, ask for a publisher page, catalogue entry or copy of the source. Phrase the request around the missing evidence. Do not describe a reference flag as a finding about authorship, intent or how a manuscript was produced.
Do not judge a reference by how plausible it looks
A familiar journal, an appropriate title and polished formatting are things to compare with a record, not a replacement for one. Research on citation validity examines unreliable reference lists. For a practical check, put the cited title alongside the title on the publication page and compare the author list, year and venue as well.
If you received a source suggestion while drafting, save it as a lead until you have opened the publication. Keep a note of where you found the actual work, rather than retaining only the suggested reference text. When a co-author supplies a citation, ask which version they read and which passage they intend it to support. Use the same process regardless of how the entry entered your bibliography.
Open identifiers and compare their destinations
Resolve the DOI, if present, and inspect the destination. A DOI identifies an object, as the DOI Foundation explains. Compare that object’s title and authors with your reference. A link that opens successfully has not completed the identity check; you still need to establish that it reaches the intended publication.
If the destination differs, keep the original entry while investigating. Search the cited title independently and compare candidate records. Do not substitute the destination’s title into your bibliography simply to make the DOI match: first decide which work your sentence needs. Record the identifier discrepancy and the source used for the eventual correction.
Search beyond an exact title failure
Try a main title with an author name, then a suitable library or subject catalogue. Crossref exposes publisher-deposited bibliographic records through its metadata API. An empty result describes that search; it does not establish absence from every possible publication collection. Keep track of which query and source you used.
For a book, check the edition rather than searching only as if it were a journal article. The ISBN Agency explains edition and format identity. For a non-English work, retain the original title and search with it where possible; CiNii’s metadata documentation gives an example of language-specific bibliographic metadata. Ask your library for help with unfamiliar catalogues instead of guessing.
Keep retractions and claim support separate
Once you have a matching work, check whether it has a publication notice. Crossref’s Retraction Watch guidance describes retractions and other updates. A retracted work has a publication identity and a notice to examine; review that notice instead of treating it as the same problem as a source you cannot locate.
Read the passage behind the claim in your manuscript. Ask whether you have cited the appropriate version and represented its findings fairly. Publication verification does not complete this reading task. Hallucite’s methodology states that its verdicts do not establish authorship or whether a source supports a claim. Use an evidence match as a starting point for reading.
Finish with an evidence note and a concrete action
Use short factual notes: “DOI reaches a different title”, “author list differs from publisher record”, or “no match found in the sources checked”. Attach the supporting link. For a confirmed error in your own manuscript, correct the reference and inspect the affected sentence. For an unresolved source, continue searching or choose a substantiated source that you have read.
Hallucite can help you compare a reference list with available publication evidence and inspect discrepancies; its limitations summary covers incomplete metadata, source outages and coverage gaps. Review each flag before acting on it. Save the corrected bibliography and the remaining review notes together so a co-author can see what you changed and what still needs attention.
Check a reference list against publication records
Hallucite compares references with available publication records and shows the evidence. Review flags and unresolved entries; a result does not establish authorship or claim support. Read the methodology and limitations