Compound Deception in Elite Peer Review: A Failure Mode Taxonomy of 100 Fabricated Citations at NeurIPS 2025
Samar Ansari · 2026
Read Compound Deception in Elite Peer Review: A Failure Mode Taxonomy of 100 Fabricated Citations at NeurIPS 2025 by Samar Ansari (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-05930.txt.
Description
Fake cites made by language models slipped into papers at a top AI meeting, and Samar Ansari lists how they were built to deceive. The study looks at one hundred made-up cites that showed in papers accepted at NeurIPS 2025, despite review by several expert researchers per paper. The fakes were spread across dozens of published papers.
A five group sorting puts the fakes by failure mode, led by total lies and partial rot of real traits. The key finding is that every single fake mixed several failure modes at once. Side traits, above all sense-truth and stolen IDs, layered onto lies to create a gloss of trust. That mixed build, the author argues, is why peer review missed them: each check is beaten by another layer. Most tainted papers held only one or two fakes, while a few held many.
Catalog record
| CREATOR | Samar Ansari |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-05930