In an extraordinary moment for the global scientific enterprise, Cornell University's arXiv repository—the premier open-access archive for physics, mathematics, computer science, and quantitative biology—has officially abandoned its founding principle of unmetered open submissions. Effective October 1, 2026, arXiv implemented strict author quotas, capping all individual researchers to two submissions per calendar month and no more than three active submissions pending moderation at any time.
The policy update was enacted as an emergency stopgap following a catastrophic surge in submission volume that pushed arXiv's human and technical infrastructure to the brink of collapse. Official telemetry released by the repository reveals an exponential trajectory: in September 2016, arXiv processed 9,869 submissions; in September 2024, volume reached 20,569. By September 2026, submissions exploded to an unprecedented 40,363 manuscripts in a single month—a doubling of monthly traffic in just 24 months.
The staggering influx of manuscripts generated 8,940 moderator support tickets in September alone, overwhelming the volunteer academic moderators who manually inspect preprints for basic scientific plausibility and proper categorical alignment. While arXiv's staff and volunteer community expanded incrementally, human throughput proved completely incapable of matching the accelerating volume.
Behind the mathematical explosion lies the widespread industrialization of generative AI in academic publishing. With frontier reasoning models able to generate complete LaTeX manuscripts, synthesize synthetic benchmark evaluations, and format academic bibliographies in minutes, the marginal cost of producing a seemingly credible research paper has fallen to near zero. Academic labs and rogue research swarms are deploying automated pipelines that draft, evaluate, and submit dozens of derivative or speculative papers weekly to amplify citation footprints.
arXiv leadership acknowledged that the monthly limit will inevitably impact prolific human research teams, particularly large collaborative physics consortia (such as CERN or high-energy astrophysics collaborations) that submit multiple preprints during major data release cycles. However, administrators emphasized that without an immediate rate-limiting ceiling, the integrity of the entire preprint commons was facing fatal degradation.
The two constraints function independently and strictly: an author who submits two papers in the first week of a month cannot submit another until the following month, regardless of whether those papers have already cleared moderation. Similarly, an author with three papers placed 'on hold' for content verification is barred from submitting any additional work until moderators resolve the pending queue.
The crisis at arXiv reflects a broader reckoning across the academic publishing ecosystem. Peer-reviewed journals and conference review pools (such as NeurIPS, ICML, and ICLR) have reported similar crises, with reviewer burnout reaching historic highs as human evaluators struggle to differentiate between genuine theoretical breakthroughs and sophisticated LLM hallucinations.
Looking forward, arXiv confirmed that it is actively collaborating with academic institutions and funding bodies to implement cryptographic identity verification. Long-term roadmaps include requiring verified institutional Shibboleth single sign-on or cryptographically signed ORCID credentials, alongside automated neural source-code analysis of submitted LaTeX archives to flag automated synthetic paper templates before they reach human moderation queues.