A peer-reviewed academic paper about the pirate ebook library Z-Library cites an article from the file-sharing news site TorrentFreak that does not exist, the site reported this week, along with at least nine other references it was unable to locate.
The paper, "Z-Library and Full-Text Access: Democratizing Knowledge or Undermining Copyright Ethics?", was published earlier this month in the journal New Review of Academic Librarianship and written by two librarians at Nigerian universities. TorrentFreak, which broke the story, said the lead author had shared a copy of the work with it.
Z-Library is one of the largest pirate ebook libraries on the internet, offering millions of books and papers for free. Publishers have spent years trying to take it down through takedown notices and blocking orders, and United States authorities indicted its alleged operators. That tension is the paper's subject, and the paper itself does not pick a side.
One reference stood out. In a section on enforcement, the authors write that major academic publishers including Elsevier and Springer Nature have taken similar legal action, and cite a 2021 TorrentFreak article listed as "Publishers continue crackdown on shadow libraries", linked to the site's homepage. TorrentFreak says it has never published an article with that title.
The site reported that it had covered a UK blocking order obtained by Elsevier and Springer Nature against Sci-Hub, and the temporary suspension of several Z-Library domains over copyright infringement claims, but was not aware of the particular legal action the paper describes.
Other claims did not fit either. The paper's account of the Z-Library criminal case states that the US Department of Justice "arrested individuals" in 2022. The department's own press release says that two alleged operators were arrested in Argentina "at the request of the United States", and the title the paper gives for that press release does not exist either.
Further entries could not be traced. A 2019 Harvard Journal of Law & Technology article credited to an intellectual property law professor is missing from the table of contents of the issue cited, whose cited pages are taken up by two articles on other topics. A 2019 College & Research Libraries article is credited to an author who has nothing in that issue. In total TorrentFreak had trouble finding at least ten of the paper's 44 references, as well as others that did not necessarily back the claims they were attached to.
The acknowledgements may explain where the mystery sources came from. "We confirm that Consensus AI was used in sifting through large volume of literature and extract relevant information," the authors write, adding that ChatGPT was used to improve the language of parts of the paper. AI chatbots are known to invent references, often with plausible titles credited to real experts in the field.
TorrentFreak also ran the paper's abstract and conclusion through the AI detection tool Pangram, which reported for both that "100% of the text is likely AI-generated". The site noted that such tools are not perfect and can produce false positives.
The paper's own warning is unambiguous. "The widespread use of Z-Library poses significant challenges to the sustainability of academic publishing. Publishers rely on subscription fees and paywalls to fund the peer review, editing, and distribution of scholarly works," it reads.
The publisher, Taylor & Francis, told TorrentFreak: "I can confirm that this has been referred to our Editorial and Publishing Ethics & Integrity teams, who will investigate the concerns you have raised." It will not comment further while the investigation is open. The lead author did not answer follow-up questions about the references. Whether the paper will be corrected or retracted has yet to be seen.
The paper's central claim can usefully be tested against the paper. Subscription fees and paywalls fund peer review, editing and distribution. This paper passed through all three, at a commercial publisher, and arrived with at least ten references that cannot be found.
That gives a measurable figure for what the funding bought: at most thirty-four locatable sources out of forty-four. Our analysts put the yield at no better than seventy-seven per cent, and note that the shortfall cannot be blamed on shadow libraries, because the missing works are not available anywhere, including there.
The Department of Unnecessary Research regards this as the strongest form of evidence a paper can offer, which is to demonstrate its thesis in the bibliography rather than argue it in the text. No control group was required. The sustainability of academic publishing was measured directly, in references, and came out at ten short.
There is also a point of copyright law that enforcement bodies may wish to consider. A pirate library can only distribute works that were written.
Ten of this paper's forty-four sources therefore enjoy perfect protection, having never existed, and are the only entries in its bibliography that nobody can possibly download without permission.