Researchers debate peer review anonymity rules and whether public reviews can trace AI slop papers

A discussion on ICLR's review reform has branched into two topics: how reviewer anonymity varies by conference and role, and whether open reviews can be used to track mass-submitted low-quality AI papers. Scholars in the discussion broadly agree that, at least at senior review levels like PC, submitters' real identities are often visible.

Confirmed

undohubris noted that ICLR publishes even rejected papers and their reviews, making it easy to track who is bulk-submitting slop, and suggested other conferences follow suit.
Anka Reuel, drawing on her own experience, said that on OpenReview for some conferences she has indeed seen reviewers' and authors' real names, depending on reviewer role and conference configuration; to her knowledge, at least the PC level can see the real names of everyone involved in a submission.
She later corrected herself: not all conferences let senior reviewers see submitters' real names, but most conferences' PCs can; she believes this doesn't change the judgment about ICLR.
For example, NeurIPS 2026 officially states its review process is double-blind at the AC level but not at the SAC and PC levels. This clarification was in response to Phillip Isola's question about whether senior roles should also be double-blind.

Why it matters

If ICLR-style open reviews (including rejections) were adopted by other conferences, it would significantly raise the cost and traceability of bulk-submitting low-quality AI-generated papers.
The inconsistent anonymity policies across major conferences (double-blind ACs vs non-blind SAC/PC) provide a concrete reference point for debates on review fairness and accountability, showing that the practical boundaries of "double-blind" vary by role.

2026-09-20 ~ 2026-09-21 · 5 related posts

Primary sources

ICLR review anonymity debate: most conferences still let senior reviewers and PCs see real names — AnkaReuel · 2026-09-21 01:43 (UTC+08:00)
Researchers debate whether ICLR's public reviews can expose AI slop paper submitters — undo_hubris · 2026-09-20 14:15 (UTC+08:00)
NeurIPS 2026: double-blind at AC level, but not for SACs and PCs — AnkaReuel · 2026-09-20 15:12 (UTC+08:00)

[source] Researchers debate whether ICLR's public reviews can expose AI slop paper submitters — undo_hubris · 2026-09-20
Reviewer says real names of reviewers and authors are often visible on OpenReview — AnkaReuel · 2026-09-20
[source] NeurIPS 2026: double-blind at AC level, but not for SACs and PCs — AnkaReuel · 2026-09-20
[source] ICLR review anonymity debate: most conferences still let senior reviewers and PCs see real names — AnkaReuel · 2026-09-21

1 near-duplicate retellings: AnkaReuel