arXiv Just Put AI Paper Spam on a Ration: Two Papers a Month, Max
October 4, 2026 — Ithaca, NY
The world’s main preprint server for AI research is rationing uploads. arXiv now limits each submitter to two papers per calendar month, with no more than three active submissions at any time — a direct response to a flood of low-quality, AI-generated papers that has overwhelmed its volunteer moderators.
What happened
arXiv announced the updated rate limit policy on October 1, 2026, and it took effect immediately. The rule applies across every subject category, not just computer science. Rejected papers count against the monthly quota, and papers withdrawn before announcement don’t. The limit is per submitter — co-authors aren’t charged for a paper someone else uploads.
Why the cap was needed
The numbers tell the story. arXiv received a record 40,363 submissions in September 2026, nearly double the 20,569 it took in September 2024, and more than four times the September 2016 figure. The computer science AI category alone grew more than sixfold in those two years. September generated roughly 9,000 support tickets for a moderation team of about 300 volunteers.
Moderators report a surge in “thin papers of narrow scope” and “salami papers” — single works sliced into several submissions to multiply the count. “A relatively small proportion of authors are submitting a large number of low-quality papers and consuming a disproportionate fraction of the moderators’ time,” said Thomas Dietterich, Distinguished Professor Emeritus at Oregon State University and chair of arXiv’s Editorial Advisory Council. He said that’s unfair to authors of quality papers, whose work can be delayed days or weeks as a result.
arXiv’s own post names the driver plainly: AI tools make it trivial to generate plausible-looking papers in minutes, making it easy “to flood arXiv and other repositories with these low-value papers.” The policy still permits AI as a research tool, as long as its use is disclosed and the work meets arXiv’s standards of scholarly interest.
Why it matters
This is AI’s credibility loop closing. The same technology writing the papers is straining the very pipeline that vets AI research — the venue where nearly every major AI breakthrough first appears. For readers, it’s a reminder to check whether that headline “new study” ever passed real peer review. For researchers, expect a more selective rhythm: two shots a month, so drafts get curated, not firehosed. The cap is framed as a stopgap while arXiv updates its rules and moderation tools — expect more guardrails, not fewer.
FAQ
What is arXiv?
arXiv is the free, open-access preprint repository where most physics, math, and AI research is posted before formal peer review. Cornell’s arXiv has run it for over three decades.
Does the limit apply per author or per account?
Per submitter — the person who presses the upload button. Co-authors on someone else’s upload don’t spend their own quota.
Are AI-assisted papers banned?
No. arXiv permits AI as a tool if its use is disclosed and the research meets its standards for scholarly interest and advancement of the field.
How long will the cap last?
arXiv hasn’t said. It calls the policy temporary and says it will monitor the effect and adjust.
Sources: arXiv blog (“Fair Moderation, Equitable Access, and AI”), Cybernews.