arXiv received a record 42,152 research paper submissions in October 2026, the first month of its new rate limit of two submissions per submitter per calendar month, according to Times Higher Education (reported 11 November 2026). The volume that prompted the cap, 40,363 submissions in September, was exceeded rather than reduced, which tells research offices that per-author limits alone do not stop volume growth.
What arXiv changed on 1 October 2026
In a post dated 1 October 2026, arXiv announced an updated rate-limiting policy across all submitters and categories. Submitters may make up to two submissions per calendar month, with no more than three active submissions at any one time. arXiv gives the aim as distributing its volunteer moderators’ time equitably across authors and responding to “a sharp increase in inappropriate submissions”, and it calls the policy “a stopgap” while it works out best practice for authors using advanced AI tools.
arXiv’s FAQ sets out how the limits are counted:
- The limit counts submissions, not announced papers, so rejected papers use up the monthly allowance.
- Papers deleted before announcement are not counted.
- Papers held from the previous month do not count toward the next month’s limit but do count toward the three active submissions.
- The limits apply only to the submitter, so co-authors are not charged against their own limits.
arXiv says it permits authors to use AI as a tool in support of research as long as the use is disclosed and the work meets its scholarly standards. It also cites about 9,000 support tickets generated for moderators in the lead-up to the change. 404 Media’s 5 October report quotes arXiv’s computing-section chair Thomas Dietterich saying that “a relatively small proportion of authors are submitting a large number of low-quality papers”. Rate limiting already existed, but it was left to individual moderators’ discretion.
What the October figures show
Times Higher Education, in an article by Frances Jones published on 11 November 2026 and updated on 12 November, reports:
- 40,363 submissions in September 2026, described as unmanageable.
- A record 42,152 research paper submissions in October 2026, despite the cap.
- 9,465 mathematics papers in October, 6.5 per cent more than in September.
The same article reports an unpublished analysis by Originality.ai founder Jonathan Gillham suggesting that some authors may be sharing submission duties with co-authors, which would sidestep a limit that applies only to the submitter. That is one analyst’s early reading, not a finding by arXiv, and arXiv’s own post does not describe how it intends to detect or respond to such behaviour.
Why a rising count does not by itself mean the cap failed
Two months of data cannot separate the effect of the cap from underlying growth. A cap on each submitter limits what one account can post; it does not limit the number of accounts, the number of co-authored papers or the overall pool of researchers using AI tools to draft. The figures reported so far also do not show how many of the October submissions came from submitters who hit the limit, or how many were rejected. Those are the numbers that would show whether the policy is changing behaviour.
Other publishers are setting caps too
arXiv is not alone in limiting submissions per author. The American Journal of Political Science announced on 4 November 2026, effective immediately, that each person may author or coauthor at most three submissions in any rolling twelve-month period. Unlike arXiv’s submitter-only limit, the AJPS limit applies to every coauthor on a team, creating multiple Editorial Manager profiles to get around it is prohibited, and one waiver a year may be requested. The journal says its submissions rose 19 per cent from 2025 to 2026 and that AI tools are helping some scholars produce many more papers. A journal cap manages editor and reviewer workload; a preprint server cap manages moderation workload before any peer review.
What this means for authors and research offices
- Plan submission timing. arXiv itself asks authors to plan carefully around the limits, including around conference deadlines. A lab posting several papers at once now needs to decide which submitter posts which paper.
- Treat the submitter as a recorded role. Because the limit attaches to the submitter and not to every author, agreeing who submits is now a practical question for collaborations. Spreading submissions across co-authors is the very behaviour Times Higher Education reports analysts are looking for, so groups should make the choice for ordinary reasons, such as who holds the endorsement and the final files, and not to avoid a limit.
- Check disclosure and quality. arXiv’s rule that AI use be disclosed still applies, alongside its earlier penalties for unchecked AI content (see our report on arXiv’s one-year ban for unchecked AI content). Internal review before posting reduces the risk of a rejection that still uses up the monthly allowance.
- Review internal guidance. Research offices that advise on preprints may want to update guidance on arXiv’s rate limit and on how it interacts with arXiv endorsement, which we covered in arXiv’s 2026 policy overhaul. Our arXiv submission guide covers the mechanics of a submission.
arXiv says it will keep monitoring the effect of the update and may change it.
Sources
- Times Higher Education, “Submissions to preprint server keep rising despite author cap” (11 November 2026, updated 12 November): https://www.timeshighereducation.com/news/submissions-preprint-server-keep-rising-despite-author-cap
- arXiv blog, “Updated rate limit policy” (1 October 2026): https://blog.arxiv.org/2026/10/01/updated-rate-limit-policy/
- 404 Media, “arXiv is rate limiting submissions because it can’t keep up with AI slop” (5 October 2026): https://404media.co/arxiv-is-rate-limiting-submissions-because-it-cant-keep-up-with-ai-slop
- American Journal of Political Science, new submission cap policy (4 November 2026): https://ajps.org/2026/11/04/new-ajps-submission-cap-policy/
