Scientific Archive Limits Paper Submissions to Stop Flood of Artificial Intelligence Content

2026-10-03 economy

Ithaca, Sunday, 4 October 2026.
Facing a record 40,363 monthly submissions, scientific repository arXiv capped uploads to two papers per author monthly to curb low-quality, AI-generated research overwhelming its volunteer moderators.

Implementation of Strict Submission Caps

Effective October 1, 2026, the open-access scientific archive arXiv enforced a hard limit of two paper submissions per calendar month for each submitting account [1][2]. This policy also restricts users to no more than three papers simultaneously under moderation, with rejected submissions counting toward the monthly quota [1][5]. The measure was implemented as a temporary stopgap to manage an overwhelming influx of low-quality and AI-generated content straining the platform’s moderation infrastructure [2][8]. arXiv officials clarified that the limit applies specifically to the individual submitter rather than co-authors, allowing collaborative groups to coordinate uploads across multiple accounts [2][5]. This structural change marks a significant shift from previous norms, aiming to restore balance between submission volume and the capacity of approximately 300 volunteer moderators [1][5].

Surge in Volume and AI-Driven Growth

The decision follows a record-breaking month in September 2026, during which arXiv received 40,363 submissions, nearly double the 20,569 submissions recorded in September 2024 [2][8]. This recent volume represents a growth rate of 96.232 percent over the two-year period, driven largely by the artificial intelligence category [2][5]. Over the past decade, submission volume has increased by 308.988 percent, rising from 9,869 submissions in September 2016 to the current record highs [5][6]. The AI category specifically grew more than sixfold between 2024 and 2026, contributing significantly to the moderation backlog [1][2]. In September 2026 alone, the surge generated nearly 9,000 support tickets for staff and moderators to process [2][6].

Economic and Academic Implications

For enterprise research and development teams, the restriction could slow the public dissemination rate of breakthrough machine learning models and signal upcoming regulatory standards across scientific publishing platforms [1]. Thomas Dietterich, chair of arXiv’s Editorial Advisory Committee, noted that a small proportion of authors submitting large volumes of low-quality papers are consuming a disproportionate amount of moderator time [1][5]. This bottleneck creates delays for serious researchers, whose papers may wait days or weeks longer due to the congestion [5][8]. Critics argue that limiting submissions does not limit paper production, potentially pushing priority claims into private correspondence and making the community less open [4]. Some researchers fear the policy could render arXiv obsolete for time-sensitive fields where establishing priority via public timestamp is critical [6].

Future Monitoring and Strategic Adjustments

arXiv staff will continuously monitor the impact of these new submission policies on authors, readers, and the overall repository, with the intention to modify the rules as necessary in the future [2]. The platform describes the current submission rate limit as a temporary measure, though no specific timeline for relaxation or removal has been provided [1][5]. In May 2026, arXiv established a one-strike ban policy for papers containing AI hallucinations, indicating a broader strategy to manage AI-generated content quality [1]. Additionally, a $17.2 million donation announced on September 23, 2026, is intended in part to develop technology for managing AI-generated content over the next three to five years [1]. As the scientific community adapts, labs accustomed to centralized uploads will need to rethink their submission strategies to comply with the new per-account restrictions [1][8].

Sources

Artificial Intelligence Academic Research