Citation Checker

The Citation Checker verifies whether the citations in a document are real and accurately described. When enabled, Copyleaks finds every reference in the submitted text, identifies the work it points to, and verifies it against a trusted source, so you can catch fabricated, misattributed, or incorrectly dated citations. It covers two kinds of references:

Academic references (papers, journal articles) are verified against the Copyleaks academic citation index.
Non-academic references (web pages, documentation, blogs, encyclopedias) are verified by fetching the cited URL and comparing the live page to what was cited.

Benefits and Use Cases

AI-generated and low-quality writing often includes citations that look real but are not. The Citation Checker surfaces those automatically, helping you:

Catch fabricated or hallucinated citations - references to works that do not exist or cannot be found.
Spot misattribution - wrong authors, wrong year, or a title that does not match the cited source.
Verify bibliographies at scale - check every citation in a submission without manual lookup.

Common use cases:

Academic integrity - flag invented or misquoted sources in student work.
Student writing support - help students improve their citations by catching missing, incorrect, or mismatched references before they submit.
Publishing and research - confirm a manuscript’s references are genuine and correctly described before publication.
Content review - assess how trustworthy the cited material in a submitted document is.

Validation Results

For each detected reference you get what Copyleaks parsed from the citation, the corroborating source it found, and a per-field breakdown of which details matched (title, authors, year). Results are delivered in two places:

Completed webhook - a referencesValidation.summary with the headline counts: how many references were found and how many were fully validated.
Crawled version - the full per-reference results, including the parsed fields and the corroborating sources.

Citations Cross-Check

Citations Cross-Check is an optional part of the Citation Checker. Validation tells you whether each reference is real. Cross-check tells you whether the document actually uses its reference list correctly, by comparing the in-text citations, such as (Smith, 2020) or [3], with the entries in the reference list. It works in both directions:

Citation to reference - every in-text citation should point to an entry in the reference list. A citation with no entry is flagged as not matched.
Reference to citation - every entry in the reference list should be cited somewhere in the text. An entry that is never cited is flagged as not cited.

When a match is close but not exact, for example the author is right but the year is different, the item is marked as needing review instead of being flagged outright, so a human can decide. This helps you catch:

Missing references - sources cited in the text that never made it into the reference list.
Padded reference lists - entries listed but never used in the text.
Wrong years and mixed-up sources - citations that point to the right author but the wrong work.

Turn it on with references.crossCheck. It needs the parsed reference list, so it turns on reference validation automatically. The results are delivered next to the validation results: a citationsCrossCheck.summary in the completed webhook, and the full per-citation results in the crawled version.

Cross-check supports author-date citations (APA, Harvard, MLA author-page) and numbered citations (IEEE, Vancouver). Footnote-style documents are detected but not cross-checked, so nothing is flagged. It works in English, Spanish, Portuguese, French, German, Italian, Danish, and Russian, with partial support for Dutch. In Chinese and Japanese documents the reference list is still validated, but in-text citations are not detected and nothing is flagged.

Next steps