TorrentFreak reported on October 24, 2026 that a peer-reviewed paper about the pirate ebook library Z-Library, published earlier in October in the journal New Review of Academic Librarianship, cites a TorrentFreak article that was never published. The paper, written by two librarians at Nigerian universities and published by Taylor and Francis, is titled "Z-Library and Full-Text Access: Democratizing Knowledge or Undermining Copyright Ethics?" When TorrentFreak checked the full reference list, it could not locate at least 10 of the paper's 44 references.
T
he direct answer: a peer-reviewed paper published in October 2026 cited at least 10 references that cannot be found, including a TorrentFreak article that was never published, and its publisher has opened an ethics investigation into the matter.
The phantom shelf
The fake TorrentFreak article is the cleanest catch, because the outlet checked its own archive. In a section on enforcement against Z-Library, the paper claims major academic publishers, including Elsevier and Springer Nature, took similar legal action, citing a 2021 TorrentFreak article. The reference section lists it as "Publishers continue crackdown on shadow libraries" and links to the TorrentFreak homepage. TorrentFreak says it never published an article with that title, and it is not aware of that particular legal action against Z-Library.
| Cited source | What checking found |
|---|---|
| TorrentFreak article, 2021, "Publishers continue crackdown on shadow libraries" | No such article in TorrentFreak's archive; the link points to the homepage |
| US Department of Justice press release on the Z-Library case | The title given in the paper matches no real release; the DOJ release says the two alleged operators were arrested in Argentina at the request of the United States |
| Daniel Gervais, Harvard Journal of Law and Technology, 2019, "The future of intellectual property in the shadow of shadow libraries" | The cited issue's table of contents does not include it; the cited pages belong to two articles on other topics |
| Carol Tenopir and colleagues, College and Research Libraries, 2019 | The cited issue has no article by Tenopir; she co-authored a 2019 article on a similar topic in a different journal under a different title |
| Article credited to Björn Brembs, 2019; articles credited to Smith, Piper, Okike, and Aronson (year not stated in the source) | Could not be located |
| Kaspersky report, "The hidden dangers of shadow libraries" | Could not be located |
The pattern matters more than any single entry. Every phantom citation is plausible: real outlets, real experts, fitting context. TorrentFreak also notes that other references in the paper do not necessarily back the claims they are attached to. The paper's account of the Z-Library criminal case, for example, states that the DOJ "arrested individuals" in 2022, which does not match the DOJ's own release.
How the paper says AI was used
The paper's acknowledgements say Consensus AI was used to sift through a large volume of literature and extract relevant information, and that ChatGPT was used to improve the language of parts of the paper. TorrentFreak ran the paper's abstract and conclusion through the AI detection tool Pangram, which reported that "100% of the text is likely AI-generated" for both sections.
That detector result needs its caveat stated plainly, because TorrentFreak states it too: AI detection tools are not perfect and can produce false positives. A Pangram score does not prove AI wrote the phantom references. What it does, combined with the acknowledgements, is establish that AI tools were part of the research and writing process of a paper whose reference list contains at least 10 entries that cannot be found. AI chatbots are known to invent references, often with plausible titles credited to real experts in the field. That is the leading explanation, not a proven fact.
Why peer review missed it
A reference list is exactly the kind of thing peer review is supposed to catch, and exactly the kind of thing it rarely checks line by line. Reviewers read for argument, method, and contribution. They do not, as a rule, open all 44 references and verify each title against the cited journal's table of contents. A fabricated citation that names a real expert, a real journal, and a plausible title sails through a review process built on trust.
The irony in this case is sharp. The paper explicitly warns that Z-Library hurts publishers' revenue, which funds peer review and editing. It does not mention that peer reviewers generally are not paid for their work, an issue central to a recent lawsuit against major academic publishers including Taylor and Francis. And the episode lands weeks after our examination of Harvard's study on AI-assisted peer review, which found AI reviewers preferred AI-written work and human editors still caught different problems. The review system is now grading AI-assisted work with tools and habits built for a pre-AI world, and this paper is what slips through the gap.
How to check a citation in 60 seconds
Readers do not need special tools to catch most phantom citations. Search the exact title in quotation marks. If the only hit is the paper citing it, that is a warning sign. Open the cited journal issue's table of contents and look for the article, rather than trusting the citation. Check that the DOI resolves to the claimed paper, not to a journal homepage. Any one of these checks would have caught the TorrentFreak entry, because the title returns nothing and the link goes to a homepage, not an article.
What happens next
Taylor and Francis told TorrentFreak the matter has been referred to its Editorial and Publishing Ethics and Integrity teams, who will investigate. The publisher will not comment further while the investigation is ongoing, and whether the paper will be corrected or retracted has yet to be decided. The lead author initially shared a copy of the paper with TorrentFreak but did not answer follow-up questions about the references. TorrentFreak also asked Gervais, Brembs, and Tenopir whether they wrote the papers credited to them, and none responded.
The honest summary is narrow and damning at the same time. This is one paper, not proof that peer review is broken everywhere. But it is proof of a specific failure mode: a peer-reviewed journal published a reference list that nobody verified, in a paper that discloses AI assistance, and the errors were found by the outlet whose name was borrowed, not by the review process. Until journals check references the way they check statistics, phantom citations will keep passing.
By ByteHerald Staff. TorrentFreak reported on October 24, 2026 that a peer-reviewed paper on Z-Library, published in October 2026 in New Review of Academic Librarianship by Taylor and Francis, cited a TorrentFreak article that was never published, and that at least 10 of the paper's 44 references could not be located. The paper's acknowledgements state that Consensus AI was used to sift literature and ChatGPT was used to improve language, and TorrentFreak's Pangram check reported the abstract and conclusion as 100% likely AI-generated, with the caveat that detection tools can produce false positives. Taylor and Francis confirmed on October 24, 2026 that its Editorial and Publishing Ethics and Integrity teams are investigating, with correction or retraction undecided.
Comments
Post a Comment