AI Citation Hallucination Statistics (2026 Data)

In the 30 days ending July 8, 2026, people submitted 47,098 references to the CiteMe Reference Checker across 2,149 checks. Every reference was compared against real academic databases. Here is what that verification produced:

• 27.2% of all checked references (12,803 of 47,098) carried the signature of a fabricated — "hallucinated" — citation.
• 84.4% of full bibliographies (5 or more references) contained at least one reference flagged as likely fabricated.
• 90.5% of full bibliographies contained at least one reference that could not be verified at all.
• Only 53.9% of references matched a real, findable work cleanly; 12.2% matched a real work but with wrong metadata (year, authors, or journal); 34.0% could not be matched to any record.
• The median checked bibliography had 17 references.

These statistics are first-party data from the CiteMe Reference Checker and Citation Checker. Users paste a reference list (or upload a document), and each entry is checked against scholarly databases — OpenAlex, Crossref, Semantic Scholar, PubMed, Europe PMC, and others — by title, authors, year, journal, DOI, and other identifiers. The numbers above aggregate every check completed in the 30-day window ending July 8, 2026, counted from server-side verification logs rather than optional browser analytics, so consent settings and ad blockers do not skew them.

All figures are aggregates. No individual bibliography, reference, document, or user is identifiable from this data, and cells below a minimum sample threshold are suppressed in our reporting pipeline.

A reference is flagged with a fabrication signature only when it fails verification in a specific way: it is well-formed — plausible authors, a real journal name, a complete DOI — yet its identifiers resolve to nothing and no acceptable title-author-year match exists in any connected database. That combination of confident formatting and zero underlying record is the fingerprint of a language-model-invented source.

The flag is deliberately conservative. A reference that is merely hard to find — an old book, a regional journal, a government report outside the indexed databases — lands in the broader "unverified" group (34.0%), not the fabricated group (27.2%). The true fabrication rate in submitted bibliographies is therefore at least as high as the flagged rate, not lower.

The headline number: 27.2% of the 47,098 references checked in this window — roughly one in four — showed the fabrication signature. This is not a formatting problem. A formatting error still points to a real paper. These references point at nothing.

The bibliography-level view is more striking than the reference-level view. Of the 1,553 checks that involved a full bibliography (5 or more references), 84.4% contained at least one reference flagged as likely fabricated, and 90.5% contained at least one reference that could not be verified. A clean submitted bibliography — every reference verified — is now the exception in this population, not the norm.

If you want to understand how these fake references are produced and how to recognise one by eye, our companion guide on spotting AI-hallucinated citations covers the mechanics; this page is the measurement.

Peer-reviewed studies that asked chatbots to produce citations and then verified them report fabrication rates in the same range or higher. Walters and Wilder (Scientific Reports, 2023) found that 55% of GPT-3.5's citations and 18% of GPT-4's citations were fabricated in a controlled test. A 2023 Cureus study of ChatGPT-generated medical content found nearly half of the references were fabricated and most of the rest contained errors. Research on human-written papers, before generative AI, already put general citation error rates at 25-54% — but those were mostly metadata mistakes pointing at real works, not invented sources.

Our data adds the missing piece: not what a model produces in a lab prompt, but what actually reaches the bibliographies people are about to submit — after whatever mix of AI drafting, manual editing, and copy-pasting produced them. At 27.2% flagged fabricated, the real-world contamination rate sits squarely inside the range the laboratory studies predicted.

You are welcome to reference this data in articles, guides, teaching materials, and research with attribution. A ready-made citation:

CiteMe (2026). AI Citation Hallucination Statistics: verification outcomes for 47,098 checked references, 30 days ending July 8, 2026. https://citeme.app/learn/citation-hallucination-statistics

The underlying aggregates are also available as a machine-readable JSON dataset (CC BY 4.0) via the Download Dataset button on the chart below. We plan to refresh these figures periodically as checking volume grows; the published date above matches the data window described on this page, and revisions will carry an updated date and a new dated dataset file.