arXiv Implements Rate Limiting on Submissions Amid Surge in AI-Generated Papers

arXiv has projected a staggering increase in submissions, from 20,569 in September 2024 to an anticipated 40,363 by September 2026. This follows a doubling of submissions over the last two years and a sixfold increase in the computer science category alone. In response to the overwhelming volume, arXiv has implemented several measures: it stopped accepting review articles and position papers in the computer science category less than a year ago, and in January, it required new authors to obtain endorsements from existing authors.

In May, arXiv announced a one-year ban on submissions deemed as 'AI slop', a term used to describe low-quality, AI-generated papers. Thomas Dietterich, chair of arXiv’s editorial advisory council, noted that a small number of authors are responsible for a disproportionate amount of low-quality submissions, which consume significant time from volunteer moderators. He emphasized the importance of these moderators, stating, "We are so grateful that they donate their time and expertise every day."

Despite these challenges, arXiv's policy allows authors to use AI as a tool in their research, provided that it is disclosed and meets the repository's standards for scholarly interest. The increase in submissions, particularly those generated by AI, has prompted arXiv to take these steps to ensure the quality of its content and the efficient use of its moderation resources. This situation follows previous coverage of arXiv's financial arrangements with publishers for content used in AI responses, highlighting ongoing tensions in the academic publishing landscape.