The research platform Arxiv is adjusting how it moderates submissions in its computer science category following what it called an "flood" of review papers and position papers - many of them generated with AI.
From now on, these types of articles must already have been accepted by a journal or conference and passed peer review before being uploaded to Arxiv.
Authors will need to provide proof of acceptance when submitting. Without that confirmation, Arxiv says the paper will likely be rejected.
Volunteer moderators struggle with AI paper surge
In recent years, submissions to Arxiv have surged, a trend the platform attributes in part to generative AI and large language models. These tools have made it easy to mass-produce certain types of papers. Many of the submitted review articles amount to little more than an “annotated bibliography” with minimal original discussion, according to Arxiv.
