A developer of an AI tool for conducting literature reviews said he and his team were “excited and honored” when a Cochrane journal had extended a “specific and individual invitation” in January 2025 to submit an article describing their system.
Kevin Kallmes, the chief executive officer at and founder of Nested Knowledge, and five of his colleagues wrote the manuscript and submitted their paper describing the procedure for using AutoLit. They included their affiliations and a note they held equity in the company. Cochrane Evidence Synthesis and Methods published the paper in October.
A few months later, the journal retracted it.
According to the June 1 retraction notice, the authors’ affiliations with the maker of the product described “constitute a breach of Cochrane’s COI policy.”
“Due to an editorial oversight, this policy breach was not identified during the editorial process, leading to the article’s acceptance and publication,” according to the notice, which adds all authors agreed to the retraction.
Kallmes told Retraction Watch the editorial staff “acted in good faith in the invitation, review and acceptance of our article.” However, he said he and his team were disappointed with the decision to retract, and worried some readers would assume the move reflected an author issue rather than a mistake by the journal.
The special issue containing the article covered AI in evidence synthesis and focused on validations, methods and tutorials in AI systems. The article has been cited twice, according to Clarivate’s Web of Science.
Cochrane discovered the issue when it realized the article lacked a conflict of interest statement, a spokesperson said. After asking the authors to provide one, they realized the paper failed to meet the journal’s specific conflict of interest policies. A representative from Cochrane said it shares editorial responsibility over the journal with Wiley, but Cochrane is the “steward of editorial standards.”
Kallmes told us the article was “clearly authored by the platform’s inventors based on the text and affiliations,” and said he asked the journal to issue a correction adding a COI statement.
Conflicts of interest are a “relatively uncommon” reason for journals to retract or correct papers, especially when COI is the only issue, according to a study published this month.
Last year, when employees of a tobacco company included their affiliation in the author list, but not in a separate declaration of interests, Toxicology Reports asked them to publish corrections reiterating their relationship with the company.
In March, the research integrity team at Cochrane Evidence Synthesis and Methods told the authors “Cochrane’s approach [to COI] is intentionally stricter than many traditional journals,” Kallmes said. His group plans to submit the article to a journal “not subject to Cochrane Library-specific COI policies,” he said.
“The goal of our article is to publish practices for using our software platform for systematic review,” Kallmes said. “What option should be available for publishing methods articles, given that inventors often own the methods they disclose?”
Nathan Schachtman, a product litigation attorney with an interest in COIs in studies, called the retraction a “huge embarrassment” for Cochrane. “Not only was there no failure to disclose commercial interests, the Cochrane folks solicited the article. And of course, Cochrane editors not only invited the article, the fully informed editors went forward with publication,” he told us.
Kallmes and his colleagues wrote in a statement: “We hope that our experience helps the research and publishing community identify a clear and proper approach and venue for platform developers such as ourselves to transparently publish methods for scientific research tools.”
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on X or Bluesky, like us on Facebook, follow us on LinkedIn, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].