July 30, 2026
Peer Review? More Like Peer Reboot
I flagged two research papers for fake authors and both were accepted as orals
AI papers with made-up authors got rewarded, and commenters are absolutely losing it
TLDR: Two papers with fake author details were reportedly accepted for oral presentation anyway, part of a wider wave of chatbot-made mistakes slipping into science papers. Commenters are split between rage and gallows humor, with many saying the system is becoming a loop of AI writing, AI judging, and humans barely checking.
The internet has officially entered its "nothing is real anymore" era. In this case, two researchers say they flagged conference papers for fake authors and bogus references, and both papers were still accepted as oral presentations — basically the academic version of getting caught using a fake ID and still being handed the VIP wristband. Their larger audit was even messier: out of 22 machine learning paper reviews, 15 contained made-up citations, fake author lists, or obvious chatbot sludge. They call it the "slop trenches," and commenters seemed to agree that the trenches are now the whole battlefield.
The strongest reaction? Pure doom. One commenter summed up the nightmare loop perfectly: papers are written by AI, reviewed by AI, and then used to train more AI. That line hit like a horror movie plot for science. Others were furious that fake references can slip through with so few consequences, with one person saying it should be treated more like plagiarism. Another blamed the old-school publishing system itself, arguing that if papers weren’t hidden behind paywalls and gatekeeping, checking whether a citation even exists would be much easier.
And then came the jokes. One commenter looked at exploding paper submission numbers and suggested a log plot, before predicting academia in 2027 will look like "Moltbook" — a gloriously cursed image of the future where everything is just feathers, spam, and collapse. The big drama underneath the memes: is this laziness, broken incentives, or proof that the quality-control system is quietly falling apart while everyone pretends to be too busy to notice?
Key Points
- •The authors say 15 of 22 ML conference submissions they reviewed in 2026 contained fabricated citations, fake author lists, or clearly LLM-generated writing.
- •Their review sample covered NeurIPS, WACV, and the TerraBytes workshop at ECCV, with reported issue rates of 55% for Caleb Robinson's assignments and 82% for Isaac Corley's.
- •The article cites multiple 2026 studies estimating large-scale growth in AI-generated or fabricated references across scientific publishing, including about 146,900 hallucinated citations in 2025.
- •It says peer review often fails to catch these problems, citing evidence from bioRxiv-to-publication tracking and an analysis of NeurIPS 2025 papers with hallucinated citations.
- •The article also cites studies and conference experiments suggesting substantial AI involvement in peer review, including Pangram's estimate that 21% of ICLR 2026 reviews were fully AI-generated and ICML 2026's policy-violation findings.