arXiv is capping researchers at 2 submissions a month after a September flood

arXiv received 40,363 submissions in September 2026, nearly double the 20,569 it logged in September 2024, and on October 1 the preprint server answered with a hard limit: each submitter may send in at most two papers per calendar month. The new rule applies across all subject categories, not to a single field, and counts the person who files the paper rather than everyone listed as an author.

The server’s volunteer moderators handled roughly 9,000 support tickets in the month that triggered the change, according to arXiv’s own account of why the rule was needed. The post, titled “Fair Moderation, Equitable Access, and AI: arXiv’s Updated Rate Limit Policy” on arXiv’s blog index, is credited by Techmeme to Kat Boboris of arXiv. It was published October 1, one day after the September volume was complete.

The two-a-month rule and the three-active-paper ceiling

The limit is set out in arXiv’s own updated rate limit policy, published October 1, 2026 and effective the same day. It allows two submissions per calendar month per submitter and, in addition, no more than three active submissions at any time. Both apply to every submitter in every category.

The counting rules carry most of the detail. Rejected papers still count toward the monthly limit, while papers deleted before announcement do not. Papers on hold from an earlier month do not consume the next month’s allocation but do count toward the three-active cap. Co-authors are not charged for a paper someone else submits: only the person who files it uses up a slot.

The effect for a lab that files many papers is arithmetic. A group that previously posted a paper whenever it finished one now has two slots a month per person doing the uploading, and a third paper in the same month would have to wait for the calendar to turn, since a rejected submission still uses a slot and a held one still counts against the three-active ceiling. That last rule, the per-submitter count, is the one that drew early comment. Peter Coles, who writes the blog In the Dark, noted on his blog that the limit is per submitter, not per paper author, and suggested that large collaborations could get around it by rotating which co-author handles the upload.

The September numbers behind the change

arXiv’s post puts the September total at 40,363 submissions, and Techmeme’s summary of the announcement pairs it with 20,569 for September 2024. The growth is attributed largely to AI, which, in the post’s account, made it easy to produce papers at a pace that moderation could not match. The increase brought about 9,000 support tickets and what the post describes as more low-quality submissions, including “salami” papers, in which one piece of work is split into several submissions, and thin papers of narrow scope.

The post frames the problem as concentrated rather than general. Thomas Dietterich, chair of arXiv’s Editorial Advisory Council, said that “a relatively small proportion of authors are submitting a large number of low-quality papers” and consuming a disproportionate share of moderator time. Cybernews quoted the same line and reported that arXiv calls the limit a stopgap while it develops better moderation tools and guidelines for AI-assisted research.

The policy’s stated aim, in the post’s words as relayed by Cybernews, is to distribute moderator resources equitably and to encourage quality over quantity. The cap is therefore aimed at a pattern of heavy submitting, but it is written as a flat rule rather than a threshold triggered by behavior, so the ordinary researcher who files three solid papers in a month meets the same limit as the account flooding a category with templated ones.

Reactions from Tao, Coles and conference organizers

Mathematician Terence Tao, writing on his blog the day of the announcement, summarized the policy as one that restricts submissions to at most two a month for each submitter, adopted in response to exponential growth in submissions. The comments under his post, he noted, show sharp disagreement over whether a cap fixes quality or only pushes authors toward other venues.

Cybernews adds that major conferences including CVPR, ICCV and NeurIPS have introduced per-author submission caps of their own for similar reasons, which places arXiv’s move inside a wider pattern: venues that once sorted papers by review are now rationing the intake itself, in arXiv’s case before any moderator has read a page of the submission.

Cybernews reports that arXiv calls the rule a stopgap, and the policy post sets no end date for the ceiling and no schedule for revisiting it. Until one arrives, the two-a-month ceiling and the three-active-submission limit apply to every submitter in every category, and a paper that moderators reject still consumes one of the two monthly slots.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.