Peer review can’t handle AI-written papers unless anonymous submissions and reviews go away

SimonGColton · x · 2026-08-04

A long blog post argues that AI-written papers now look good enough to fool qualified scientists, so the real fix for peer review is structural: stop using anonymous submissions and anonymous reviews.

The author says large conferences such as NeurIPS, CVPR, and AAAI are especially exposed because they handle huge volumes of submissions under double-blind review, which makes it easier for polished but potentially AI-generated papers to slip through. The post frames this as a publishing-system problem rather than just a detection problem.

Original post →

More from Research

Research channel →