Arxiv is tightening how it handles certain computer science submissions after a surge of review papers and position papers, many of them generated with AI. The platform says these article types must now clear an outside quality gate before they can be uploaded.
What Arxiv Is Changing
From now on, review papers and position papers in Arxiv’s computer science category must already have been accepted by a journal or conference. They also must have passed peer review before submission to the platform.
Authors will need to provide proof of acceptance when they upload these papers. If they cannot provide that confirmation, Arxiv says the submission will likely be rejected.
The change does not apply to regular research papers. The source also states that regular research papers exploring the social impacts of technology are unaffected.
Why AI-Generated Review Papers Became A Moderation Problem
Arxiv links the pressure on its submission system partly to generative AI and large language models. In recent years, submissions have surged, and these tools have made it easier to mass-produce certain kinds of academic writing.
The platform specifically points to review articles that contain little original analysis. According to Arxiv, many submitted review papers amount to little more than an “annotated bibliography” with minimal original discussion.
That matters because review papers and position papers can look substantial while adding little new value. A review article can collect existing work, summarize it, and present it in a formal academic format. When that process is automated or mass-produced, moderators face more material to screen without a matching increase in useful research contribution.
The source describes these papers as one of the easiest outputs for AI “paper factories” to produce. Arxiv says such papers were once rare and often high quality, but the volume and character of recent submissions have changed the moderation challenge.
Peer Review Becomes The Practical Filter
Arxiv says the policy is meant to reduce the burden on volunteer moderators. The platform’s core mission, as described in the source, is to provide fast, open access to research. The moderation change is intended to help moderators focus on that role.
The platform also says it does not have the resources to conduct in-depth quality checks itself. That is why external peer review is becoming a more important filter for these article types.
In practice, the new requirement shifts part of the screening burden away from Arxiv and toward journals and conferences. A review paper or position paper can still appear on Arxiv, but only after another venue has accepted it and peer review has taken place.
For authors, the immediate implication is simple:
- Review papers in computer science need prior journal or conference acceptance.
- Position papers in computer science need the same kind of acceptance.
- Authors must provide proof when submitting.
- Without proof, Arxiv says rejection is likely.
Not A New Rule, But A Stricter Line
Arxiv says the policy is technically not new. Review papers and position papers were never officially part of its accepted content types. Moderators had previously allowed them at their discretion.
The difference now is enforcement. Instead of relying mainly on moderator discretion for these papers, Arxiv is requiring evidence that the work has already passed through a peer-reviewed venue.
That distinction is important for understanding the scope of the change. Arxiv is not closing the door on all computer science submissions, nor is it applying the same standard to regular research papers. It is drawing a firmer boundary around review and position papers because those categories have become especially vulnerable to mass production with AI tools.
What Could Happen Next
For now, the moderation measures apply to the computer science category described in the source. But Arxiv says it may extend similar measures to additional fields if similar waves of AI-generated submissions appear elsewhere.
That makes the change a signal beyond one submission category. Arxiv is responding to a practical problem: when generative AI lowers the effort needed to produce formal-looking academic papers, open research platforms must decide how to preserve access without overwhelming moderation.
The platform’s answer, at least for computer science review papers and position papers, is to lean more heavily on external peer review. The result is a tighter upload path for certain papers, while regular research papers remain unaffected by the change described in the source.