Citation Manipulation Network on SSRN

Retraction Watch published an article about a citation manipulation scheme that Elle O’Brien, Grischa Liebel, and I helped uncover (by accident). Elle discovered a plagiarized copy of a preprint we co-authored. She noticed it because the copy cited another paper of hers, just like our original did. At first, we were confused. The title, the author names, and the email addresses had been changed, but the affiliations remained untouched. Why would anyone use fake authors when plagiarizing a paper?

The reason turned out to be added citations. The plagiarized paper contained references that were not present in our original paper and had nothing to do with its topic. When we looked into the cited papers and their authors, we found a network of fake articles on preprint servers (mainly SSRN), plagiarized from real papers (in our case published on arXiv), rewritten with GenAI, attributed to fabricated authors with real institutional affiliations, and containing additional references to inflate citation counts of specific researchers. As I told Retraction Watch, the investigation turned into “a rabbit hole.” Grischa also wrote about it on LinkedIn.

I’ve said before that I’m not a fan of celebrating citation counts and h-indices. This story is another reason to move away from citation counting as a proxy for impact. The same approach could also be used to discredit researchers by attaching low-quality citations to their work. Cleaning that up falls on the targeted researcher, who has to notice the suspicious citations in the first place, investigate them, and report each fake to the relevant preprint server.

Preprint servers are how many of us share early work. Their lightweight curation is part of what makes them useful, but tools such as Google Scholar count citations from preprints, which makes the resulting numbers easy to game.