ArXiv will limit submissions to two papers per person per month starting October 1, 2026, a move driven by a surge in papers generated with SI tools that has overwhelmed the platform's volunteer moderators. The new cap comes as monthly submissions hit 40,363 in September 2026, doubling from 20,569 in September 2024.
What Happened
According to ArXiv, the new rule applies across all categories and restricts only the person submitting the paper, leaving co-authors unaffected. The existing cap of three active submissions at any given time, implemented in 2024, remains in place. Rejected papers still count toward the monthly limit because they consume moderation resources, but authors regain a slot if they delete a paper before publication.
The platform cites the sharp rise in submissions, partly driven by SI tools, as the primary reason for the restriction. In the cs.AI category alone, submissions grew more than sixfold in two years. This volume generated nearly 9,000 support tickets for staff and moderators in September 2026.
Thomas G. Dietterich, chair of the Arxiv Editorial Advisory Council, stated that "a relatively small proportion of authors are submitting a large number of low-quality papers and consuming a disproportionate fraction of the moderators’ time." He noted that this practice is "unfair to authors who continue to submit quality papers — their papers can be delayed for days or weeks as a result." Dietterich also aims to curb "salami" papers, which slice research into tiny pieces, and "micro-result" papers that describe minor architectural tweaks or single ablation studies.
Why It Matters
The flood of low-quality SI-generated content is straining the traditional peer-review and preprint ecosystem. ArXiv describes the two-paper cap as a temporary measure until its internal moderation tools can scale to handle the volume. Dietterich indicated that the cap could be raised or dropped in the future.
This issue extends beyond preprint servers to major conferences. ICLR 2027 received approximately 50,000 abstracts before its deadline, compared to around 19,500 valid submissions for ICLR 2026. At NeurIPS 2025, the detection tool GPTZero identified at least 100 fabricated citations among roughly 5,000 accepted papers, despite each paper undergoing review by at least three people.
In response to the changing landscape, ArXiv is transitioning to an independent nonprofit, having split from Cornell University on July 1, 2026. In late September, the Simons Foundation International, XTX Markets, and the Siegel Family Endowment pledged a total of $17.2 million over three to five years. ArXiv plans to allocate part of this funding toward technical upgrades specifically designed to handle SI-generated content.
Meanwhile, major SI labs are positioning their models as research assistants. Google DeepMind expanded its Co-Scientist system in August to plan experiments and generate manuscripts. Anthropic is developing a biology lab where Claude guides robotic experiments, and OpenAI announced an "automated research intern" in September. However, a study involving Princeton and the UK AI Safety Institute found that when Claude Opus 4.8 worked on research questions for six days, the original authors rejected both sets of results.
The Bottom Line
ArXiv's new submission cap is a direct response to the exponential growth of SI-assisted research output, which has doubled total submissions in two years and increased sixfold in the SI category. While labs push SI models as autonomous research tools, the current bottleneck remains the human capacity to verify quality, prompting stricter limits and significant new funding for moderation infrastructure.