A preprint traced 2,062 IEEE conference papers to sale ads. Sixty-six have been retracted.
An updated preprint by Anna Abalkina and colleagues matched 2,062 papers in 366 IEEE conference proceedings to publication offers advertised on social media, and Science reports that 66 have been retracted so far. The count rose from 1,720 in the April version without any change in method, which tells you the figure is a floor rather than an estimate. The wider point for anyone screening literature is that conference proceedings sit outside every journal-level check a researcher would normally apply.
An updated preprint has matched 2,062 papers published in 366 IEEE conference proceedings against publication offers that were advertised for sale on social media. Science reported this week that IEEE has so far retracted 66 of them.
The study, *Opening Pandora's box: Paper mills in conference proceedings*, is by Anna Abalkina of Freie Universität Berlin with Marie Kunešová, Yagmur Ozturk and Solal Pirelli. It was posted to arXiv on 24 April 2026 and revised on 11 August 2026. It is a preprint and has not been peer reviewed.
What the study did
The team collected publication offers covering more than 4,000 unique paper titles or topics from more than 200 social media channels, then used semi-automated matching combined with human assessment to find papers actually published in IEEE proceedings whose titles corresponded to what had been advertised.
From the August version of the abstract:
- 2,062 papers across 366 IEEE conference proceedings
- up to 23.51% of the papers at a single conference
- more than 7,500 co-authors, from over 4,000 affiliations, in 56 countries
- recurring signatures including collaboration anomalies, unusually high affiliation diversity within one paper, citation manipulation, a predominance of six-author papers, and content-based irregularities
The number moved between April and August, and that is the point
The April version of the same preprint reported 1,720 papers in 286 proceedings, by more than 6,500 researchers, 3,500 affiliations, 55 countries. The August version reports 2,062 papers in 366 proceedings, 7,500 researchers, 4,000 affiliations, 56 countries. The maximum single-conference share, 23.51%, is unchanged.
Nothing in the method changed. What changed is how many advertisements the team had matched to a published paper.
Analysis
That distinction is worth holding onto, because it determines how the headline figure should be read. 2,062 is not an estimate of how many bought papers are sitting in IEEE proceedings. It is a count of the ones that could be traced back to a specific public advertisement. Every step of that chain has to survive: the ad has to have been captured, the title has to be matchable, and a human has to confirm the match. The figure is a floor, and four months of further matching moved that floor up by 342 papers.
The more useful observation for anyone screening literature is where this material lives. Conference proceedings are the part of the record with no journal-level quality instrument attached to them. A researcher assessing a journal can check its SCImago quartile, confirm it is indexed in DOAJ, look at its retraction history. None of those instruments exist for a conference proceeding. It carries the publisher's name and inherits the publisher's reputation, while inheriting almost none of the publisher's checks, because peer review at many conferences is run by whoever organised that conference.
In computer science and much of engineering, proceedings are not a lesser venue. They frequently carry the primary result. So the practical consequence is that the fields where conference publication matters most are also the fields where the standard screening moves do the least work, and a systematic reviewer in those fields is currently reliant on reading the papers themselves.
What to watch
Whether the retraction count moves materially past 66, whether the preprint clears peer review with its numbers intact, and whether any publisher other than IEEE gets examined the same way. IEEE was chosen partly because it retracts conference papers at all, which means the absence of comparable findings elsewhere would not be evidence of a cleaner record.
Sources
- Abalkina A, Kunešová M, Ozturk Y, Pirelli S. Opening Pandora's box: Paper mills in conference proceedings. arXiv:2604.22458 (v1 24 Apr 2026, v2 11 Aug 2026). https://doi.org/10.48550/arXiv.2604.22458
- Science, Hundreds of paper-mill papers peddled in ads were later published
- Retraction Watch, Weekend Reads, 15 August 2026