ArXiv versus AI slop: will capping preprint submissions work?
Nature, Published online: 15 October 2026; doi:10.1038/d41586-026-03194-w Major preprint platform tightens its rules to curb a surge of submissions, including those generated by artificial-intelligence tools.
The preprint server arXiv has implemented a limit of two papers per month for each researcher, starting on October 1st. This decision comes after a significant surge in submissions, which reached 40,000 in September alone, up from around 30,000 from March to July. The cap is intended to prevent the nearly 300 volunteer moderators from becoming overwhelmed, as they currently review about 1,000 manuscripts per day.
However, this measure is criticized by some as insufficient, as nearly 4,000 submissions are currently waiting to be posted. Some argue that the limit is reasonable, as posting two preprint articles already requires significant effort. Others believe the policy should be stricter, as it can be circumvented by sharing the submission load among co-authors.
Other preprint servers, such as SocArXiv, are also limiting submissions to one article per author per week. The issue of "AI slop" – the rapid upload of multiple papers – is a concern for many, but some argue that rejecting more articles will just lead them to other platforms. Some institutions and conferences are also implementing submission quotas to address the problem.
However, critics argue that simply rejecting more submissions will not solve the issue, as AI-generated research will simply find a new venue.
Written by urgent.news from Nature's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.