arXiv, the preprint server that has become the default publication venue for physics, mathematics and computer science, has capped how much any author can post — and pointed squarely at AI.
The new rule
Effective 1 October 2026, arXiv limits submitters to two submissions per calendar month, with a maximum of three active submissions at any time. The limit applies across all categories; rejected papers count toward it, because submissions — not published articles — are what consume moderator time. Co-authored papers count only against the person who submits them.
arXiv calls the change a "stopgap" while it works out what the new best practice should be for authors using increasingly capable AI tools, and buys time to improve its moderation tooling.
The flood, in numbers
- September 2016: 9,869 submissions. - September 2024: 20,569. - September 2026: 40,363 — a record, and roughly 9,000 support tickets for staff and volunteer moderators. - Submissions in the cs.AI category alone have grown more than sixfold in two years; other categories have doubled.
What moderators are seeing
arXiv says advanced AI tools are making research faster — and making it easy to flood repositories with low-value work. Moderators report a rise in "thin papers of narrow scope," "salami" papers that split one piece of work into several, and dense AI-written submissions.
"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. "This is unfair to authors who continue to submit quality papers — their papers can be delayed for days or weeks as a result."
Why it matters
arXiv's policy permits AI as a disclosed tool in research; what it cannot absorb is volume without value. The cap shifts the burden onto authors to curate, and it is a test case for every other open repository facing the same curve. Whether the right response is to economise on moderation or on scholarship — as one commentator put it — is now an open argument.
