Science has a peer review problem, and lately it has a peer volume problem too. arXiv, the preprint server where much of modern AI research shows up first, has officially had enough. Starting this month, everyone gets a speed limit.

Two Papers a Month, Three in Flight

In a blog post published October 1, 2026, arXiv announced that submitters are now limited to two submissions per calendar month, with no more than three active submissions at any one time. The cap applies across every category, and rejected papers still count against your monthly total, because they still took a moderator's time to review.

The numbers explain the urgency. Monthly submissions climbed from 9,869 in September 2016 to 40,363 in September 2026, doubling in just the last two years, while the cs.AI category grew sixfold in that window. September alone generated nearly 9,000 support tickets. arXiv says a relatively small share of authors is submitting a large number of low-quality papers and eating a disproportionate slice of its volunteer moderators' time.

When Writing Gets Cheap, Attention Gets Expensive

arXiv isn't banning AI. Authors can still use AI tools as long as they disclose it and the research holds up. What it's rationing is the scarce resource: human attention. When a model can draft a passable-looking paper in an afternoon, the bottleneck moves from writing to reviewing, and volunteers can't scale like GPUs do.

This is a pattern well beyond academia. Support queues, app stores, open-source maintainers, and content teams are all hitting the same wall, where generation is nearly free and verification isn't. Rate limits are a blunt tool, sure, but blunt tools exist for a reason. Sometimes the fair answer to "infinite supply" is simply "take a number."

The age of AI abundance comes with a new scarce resource, and it's the humans checking the work.

If your team is planning to use AI to churn out more content, I'd love to help you build the review workflow that keeps quality from drowning in quantity.

Source: arXiv Blog