ArXiv has capped preprint submissions at two papers per person per month starting October 1, 2026, following an overwhelming flood of artificial intelligence research that doubled monthly totals to over 40,000 in two years. The new restriction applies across all categories to the person submitting the paper, though co-authors are not affected. Rejected papers will continue to count toward the monthly limit because they consume valuable moderation time, while deleted drafts will return a slot to the author.
The surge in low-quality submissions, "salami" papers, and dense AI-generated texts has severely strained the platform's volunteer moderators and generated thousands of support tickets. Thomas G. Dietterich, chair of the Arxiv Editorial Advisory Council, noted that a small proportion of authors submitting numerous papers consume a disproportionate amount of moderator time, delaying quality research. arXiv plans to use part of a recent $17.2 million funding package from the Simons Foundation International, XTX Markets, and the Siegel Family Endowment to upgrade technical systems and handle the wave of AI content.
This policy adjustment comes as major AI labs increasingly position their models as automated research assistants capable of conducting experiments and writing manuscripts. Conferences are experiencing similar pressures, with ICLR 2027 receiving roughly 50,000 abstracts and NeurIPS grappling with fabricated citations in accepted papers. arXiv intends the new monthly submission cap as a temporary fix until scalable automated moderation tools can be deployed.
Why it matters: The submission cap highlights how generative AI output is overwhelming traditional academic infrastructure and forcing key research platforms to fundamentally rethink quality control.