arXiv Caps Monthly Submissions to Curb AI-Generated Preprint Flood

3 min read
Source: Nature
arXiv Caps Monthly Submissions to Curb AI-Generated Preprint Flood
Photo: Nature
TL;DR

arXiv has implemented a cap of two preprint submissions per researcher per month, effective 1 October 2026, to manage a surge in AI-generated manuscripts. September saw 40,000 uploads, up from 30,000 in earlier months, overwhelming its volunteer moderators. While some argue the limit is reasonable, others note it can be bypassed by co-authors. Other platforms like SocArXiv have adopted stricter weekly limits, though critics warn that rejected papers may simply migrate to other venues.

Key points

  • arXiv introduced a cap of two preprints per researcher per month starting 1 October 2026 to address a surge in AI-generated submissions.
  • Submissions to arXiv rose to 40,000 in September, up from approximately 30,000 between March and July, straining its nearly 300 volunteer moderators.
  • Moderators are currently checking about 1,000 manuscripts per day, with nearly 4,000 submissions waiting to be posted, delaying legitimate papers.
  • Siddhesh Zādéy of Columbia University argues the cap is reasonable as two monthly submissions already require substantial work, while Victor Galitski of the University of Maryland notes the limit can be circumvented by co-authors taking turns.
  • SocArXiv restricts submissions to one article per author per week, with director Philip Cohen noting that rapid successive submissions are often indicators of AI slop, though exceptions exist for dissertation chapters.
  • Natalie Khalil of Reviewer3 warns that rejecting papers at arXiv may simply shift the problem to other platforms, as AI-generated content will find alternative venues.

Background

This development follows earlier reports in October 2026 regarding OpenAI retracting three AI-generated mathematical preprints after a massive release of 722 manuscripts, which sparked academic backlash. The current arXiv cap is part of a broader trend where major AI conferences and journals are also imposing submission quotas to manage similar surges in low-quality or AI-generated content.

Why it matters

The cap aims to protect the integrity of the preprint ecosystem by preventing AI-generated 'slop' from overwhelming volunteer moderators and delaying the publication of legitimate research. However, the effectiveness of such measures is debated, as limits may be easily bypassed or simply shift the burden to other platforms, highlighting the ongoing challenge of maintaining quality in an era of rapid AI-generated content.

What to watch

Researchers and institutions will monitor the impact of the arXiv cap on submission volumes and moderator workload. Other preprint servers and journals may adjust their policies in response, potentially adopting stricter limits or alternative verification methods. The debate will continue over whether caps are sufficient or if more robust AI detection and peer review processes are needed.

Share this article

Want the full story? Read the original reporting

Read on Nature