arXiv imposes monthly submission limits to curb AI-generated paper flood
The preprint server restricts authors to two submissions per month as automated content overwhelms moderation teams and delays legitimate research.
Key points
- arXiv limits authors to two submissions per month to manage volume.
- AI-generated "slop" submissions are overwhelming human moderators.
- The policy aims to prevent delays for legitimate academic papers.
arXiv has implemented a new rate-limiting policy that restricts authors to submitting no more than two papers per month. According to research published by Socket, this measure responds to a significant surge in low-quality, AI-generated submissions that are currently overwhelming the platform’s moderation infrastructure.
The preprint server introduced these limits because the volume of automated content is delaying the review and posting of legitimate academic research. Socket reports that moderators are struggling to process the influx of "AI slop," which creates bottlenecks for genuine scholars trying to share their findings with the scientific community.
Context and Background
The decision to impose strict submission caps highlights the growing tension between open-access academic platforms and the rapid proliferation of large language model outputs. According to Socket, the current moderation workflow is insufficient to handle the sheer quantity of machine-generated text that lacks substantive value or rigorous peer review.
This policy shift sparks debate within the academic community regarding how best to apply these limits. Socket notes that there is ongoing discussion about whether the two-submission monthly cap should apply uniformly to all authors or if exceptions should be made for established researchers with high volumes of legitimate output.
Who Is Affected
The rate-limiting policy directly impacts all researchers and academics who rely on arXiv to distribute their work prior to formal journal publication. Authors who previously submitted multiple papers in a single month will now face a hard cap, requiring them to stagger their submissions over a longer period.
Moderators and the operational teams at arXiv are also affected, as the primary goal of this policy is to reduce their workload. By limiting the input volume, the platform hopes to restore efficient processing times for high-quality manuscripts that require careful human verification and categorization.
What should you do?
Researchers should adjust their submission schedules to comply with the new two-paper monthly limit to avoid rejection or delays. Institutions may need to advise their faculty on how to prioritize which manuscripts to submit first under these constraints.
Background: Large language models
Large language models are a specific type of neural network trained on text. Generative AI is the broader umbrella term for any system that creates new content, including images and code. Not all generative AI uses large language models, but most modern text generators do.
What to do and how to stay safe: arXiv
- Review your planned submission schedule for the current month to ensure you do not exceed the two-paper limit imposed by arXiv.
- Prioritize submitting high-impact or time-sensitive research first, as the new limits may delay the posting of subsequent papers.
- Monitor official arXiv announcements for any updates to the rate-limiting policy, as the platform may adjust rules based on moderation capacity.
- Verify that your manuscripts meet standard quality guidelines before submission to reduce the likelihood of being flagged as low-quality automated content.
Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
How many papers can an author submit to arXiv per month now?
Authors are limited to submitting a maximum of two papers per month under the new rate-limiting policy.
Why did arXiv implement these submission limits?
The limits were introduced to manage a flood of AI-generated submissions that are overwhelming moderators and delaying legitimate research.
Does this limit apply to all authors equally?
According to Socket, there is ongoing debate about whether the limit should apply to everyone, but the current policy restricts authors generally.
