The published record behind Reviewer3, so the claims can be checked rather than taken on trust.
Reviewer3 publishes its evaluation methodology and results in the open. The reference checker was tested on 476 citations — 299 real and 177 AI-fabricated — and correctly classified 469 of them (98.5% accuracy), catching 97.2% of the fabricated citations while correctly keeping 99.3% of the real ones. Benchmarks last updated 2026-06-15.
Two open, reproducible benchmarks. On reference verification, Reviewer3 classifies citations as real or fabricated with 98.5% accuracy across 476 citations. On review quality, ReviewBench compares more than 145,000 human and AI review comments across computer science, social science, and life science papers.
Read two blinded review comments and pick the one written by a human. A live, ongoing test of whether AI peer review is distinguishable from expert human review — you score it yourself rather than taking our word for it.
Worked examples of AI review detecting fabricated data, hidden prompts, paper-mill signatures, and methodological flaws in real published research — each one showing what was caught and how.
Upload a paper or grant and get a full review in minutes.
Get a Review