How Fake References Slipped Into a Real Journal

How Fake References Slipped Into a Real Journal

Timothy Lesaca MD

Mar 20, 2026

For decades, the credibility of academic publishing has rested on a quiet premise: that the footnotes are real. Not flawless, not immune to error, but anchored in an underlying reality that can be traced, retrieved, and verified.

What happened in a recent Springer Nature journal suggests that this premise is beginning to erode.

The episode began, as such failures often do, at the margins of the system. Jessica Waite, a hospital librarian at Royal Hallamshire Hospital in Sheffield, was asked to locate two recent references from a paper in Digestive Diseases and Sciences. This is routine work for a clinical librarian. And yet, the articles could not be found. Not behind paywalls, not under alternate titles—nowhere.

When Waite checked the journal issues where the articles supposedly appeared, she found entirely different papers.

Her suspicion widened into something more consequential: of the 14 references listed in the article, 12 did not exist.

The paper itself addressed the psychological burden following ileostomy surgery—a subject with direct implications for patient care in cases of inflammatory bowel disease, colorectal cancer, and traumatic injury. It had passed through peer review at a major publisher, Springer Nature.

And yet, its evidentiary scaffolding was, in large part, fictitious.

The author, psychologist Marie Atallah of Sutter Health, later attributed the problem to the accidental submission of an earlier draft containing placeholder citations. She acknowledged using artificial-intelligence tools during the writing process—tools known to generate “hallucinated” references—and described a drafting workflow in which such citations were sometimes retained temporarily for later verification.

Complicating matters further, Atallah cited the lingering cognitive effects of a traumatic brain injury sustained in a 2020 car accident, which she said affects executive functioning and attention to detail.

None of this explanation is implausible.

That is precisely what makes the situation so disquieting.

Even after the issue was raised, the process of correction proved unstable. A revised reference list contained additional nonexistent or mismatched citations—real articles paired with incorrect titles, journals, or authors. Only on a third attempt did a fully verifiable list of 25 sources emerge.

Throughout, the author maintained that the substance of the article remained unchanged.

Springer Nature, for its part, has said it is investigating the matter in line with Committee on Publication Ethics guidelines. The publisher has also acknowledged the broader challenge, noting that references are often inconsistently formatted and that automated detection tools risk both false negatives and false positives.

In other words, the system is not yet equipped to reliably distinguish between what exists and what merely appears to.

Academic publishing continues to operate on norms developed for a slower, more manual era—one in which generating a plausible but nonexistent citation required effort. Today, it requires only a prompt. The result is not necessarily fraud in the traditional sense, but something more diffuse: a gradual seepage of unreality into the record of knowledge.

That seepage is not easily contained. Peer reviewers are rarely tasked with verifying every reference. Editors assume a baseline of authorial accuracy. And authors, increasingly assisted by generative tools, are navigating workflows in which the line between draft and artifact has become porous.

In this context, the role of the librarian takes on renewed significance. It was not an algorithm or an editorial check that caught the problem, but a professional trained to notice when something that should exist does not. “I can see how one or two may slip through the net,” Waite observed. “But this was a preposterous number.”

The phrase lingers because it captures the scale of the shift.

When false references can pass undetected through peer review at a major journal, the question is not simply how this happened.

It is how often it is already happening.

And how much of the literature now rests on citations that lead nowhere.

The footnote, once the most reliable unit of academic truth, is becoming something else: a site where plausibility can masquerade as evidence, and where the burden of verification quietly returns to those willing to look closely enough to notice what is missing.