# ArXiv versus AI slop: will capping preprint submissions work?

Digest AI · Policy & Regulation · published 2026-10-15T00:00:00Z

Canonical: https://digestai.news/story/arxiv-versus-ai-slop-will-capping-preprint-submissions-work

## Summary

arXiv, the major preprint server, has implemented a limit of two papers per month for each researcher, effective 1 October 2026. Steinn Sigurðsson, the scientific director of arXiv, stated that this measure aims to manage a surge in submissions driven largely by low-quality content generated by artificial intelligence tools. Submission volumes rose sharply, with 40,000 preprints uploaded in September compared to roughly 30,000 between March and July. The increase has strained the site’s nearly 300 voluntary moderators, who currently review about 1,000 manuscripts daily, leaving nearly 4,000 legitimate submissions in a backlog.

Reactions to the policy are mixed. Siddhesh Zādéy, an epidemiologist at Columbia University, argued that the cap does not hinder productivity, as producing two preprints monthly is already substantial work. However, Victor Galitski, a physicist at the University of Maryland, noted that the rule can be bypassed by co-authors taking turns to submit. Philip Cohen, director of the social-sciences repository SocArXiv, which limits submissions to one per week, suggested that rapid successive uploads are often a sign of AI-generated "slop." Natalie Khalil, founder of Reviewer3, warned that rejecting these papers may simply shift the problem to other platforms rather than solving the underlying issue of AI-generated content flooding academic repositories.

## Key points

- arXiv limits submissions to two papers per researcher per month starting 1 October.
- September saw 40,000 preprints uploaded, up from about 30,000 in March to July.
- Nearly 4,000 submissions are currently waiting to be posted due to moderation backlog.

## Why it matters

The policy addresses the integrity of scientific communication by mitigating the impact of AI-generated low-quality content on peer review processes and repository accessibility for legitimate researchers.

## Sources

1. [ArXiv versus AI slop: will capping preprint submissions work?](https://nature.com/articles/d41586-026-03194-w) (Nature Machine Learning, 2026-10-15, primary source)

Part of the developing story: [ArXiv AI Submission Saga](https://digestai.news/thread/arxiv-caps-submissions-to-two-per-month-amid-ai-driven-research-surge) (2 stories)

## Cite

Digest AI, "ArXiv versus AI slop: will capping preprint submissions work?", 15 October 2026, https://digestai.news/story/arxiv-versus-ai-slop-will-capping-preprint-submissions-work

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/arxiv-versus-ai-slop-will-capping-preprint-submissions-work.json