arXiv now caps every submitter at two papers a month, as AI tools help double submissions
September brought arXiv 40,363 submissions, double the count two years earlier, and a rejected paper now uses up half your month.
Two papers a month. That's now all any one person can send arXiv, the free preprint server, under a rule that took effect on October 1, 2026. arXiv announced it the same day in a post on its blog by Kat Boboris.
It covers every category on the site. What surprised me is that a rejected paper still counts against you. The older ceiling of three active submissions at once (it's been there since 2024) stays as it is.
- Limit
- Two submissions per calendar month
- Active at once
- Three total active submissions
- Applies to
- All submissions, across all categories
- In effect
- October 1st, 2026
- September 2026
- 40,363 submissions
Why arXiv did it
The numbers doubled in two years. arXiv got 9,869 submissions in September 2016 and 20,569 in September 2024. This September it got 40,363, a record, and they brought in almost 9,000 support tickets for staff and moderators.
The post puts much of the jump down to AI tools. Moderators report thin "salami" papers, where one piece of work gets cut into several, and a marked increase in dense, AI-written papers. arXiv still lets authors use AI when they disclose it and the work meets its standards. The cs.AI category alone grew more than sixfold in two years (other categories doubled).
a relatively small proportion of authors are submitting a large number of low-quality papers
arXiv calls the cap a stopgap. It wants time to work out what good practice looks like for authors leaning on stronger AI tools, and to improve its own moderation tools.
The fine print
arXiv counts submissions rather than announced papers, since submissions are what eat moderators' time. So if you send two in October, you wait for November, whatever happens to them. Delete one before it's announced, though, and it doesn't count.
Only the person who clicks submit is limited. Co-authors aren't touched at all, which means a group can spread its papers across several submitters, and arXiv just asks them to coordinate. Got a conference deadline? Plan ahead, it says.
What researchers said
Thomas Dietterich, who chairs arXiv's Editorial Advisory Council, flagged the change to authors on X the afternoon it went up.
Attention arXiv authors. Because of the exponential increase in submissions, arXiv is limiting the number of papers that a person can submit each month to two. 1/
Some researchers read the cap as aimed at a small group of heavy submitters, and said so within hours.
Seems like a good initiative. Suppose it is primarily the first author who submits the paper, hard to imagine anyone need submit more than two within a single month. It only penalizes extreme gaming cases.
Joshua Saxe took it somewhere bigger, to how machine learning research gets done at all now.
arXiv and every other academic venue is being denial of serviced with slop submissions. It feels like every pillar of what I grew up with as a (self-taught) grant funded machine learning researcher is being disrupted. The driving research is being done privately and in secret or is published partially as marketing by closed labs. Academia is publishing research papers on black box commercial products (e.g. the OpenAI API) and academics are often marginalized unless they build up large social media and press followings. https://t.co/QhRUsTiUX3 is full of slop unless you filter by affilitation or trusted author. This obviously isn't just machine learning it's metastasizing across every discipline. Now's the time for bold proposals on the structure of the scientific enterprise.
Bojan Tunguz was bleaker. Dan Roy, posting early the next morning, gave arXiv credit for moving first.
arXiv is finished. It can't survive the AI revolution.
Conferences have been talking about this for years. ArXiv just did it.
arXiv says it'll watch how the rule lands on authors and readers and change it if it needs to. Anyone who has already sent two papers since October 1 can submit again in November.