ArXiv, one of the world’s most important repositories for research preprints, is moving aggressively to curb what moderators call AI-generated “slop,” escalating a broader academic struggle over how to use large language models without undermining trust in science.
Early concerns and groundwork
ArXiv has been grappling with AI-written, low‑quality papers for some time. It previously tightened controls by requiring many first‑time authors to secure an endorsement from an established researcher, aiming to stem a “growing number of low-quality, AI-generated papers.”1 The site’s shift to an independent nonprofit structure was also framed as a way to raise more resources to address problems like AI junk submissions.1
The new one-year ban policy emerges
The latest crackdown crystallized this week as coverage described how ArXiv will now ban authors for a year if there is “incontrovertible evidence that the authors did not check the results of LLM generation,” including obvious markers like hallucinated references or visible LLM meta‑comments in manuscripts.2 One report summarized the move bluntly: “Send the arXiv AI-generated slop, get a yearlong vacation from submissions.”3
Computer scientist Thomas Dietterich, who chairs ArXiv’s computer science section, explained that such evidence means moderators “can’t trust anything in the paper,” triggering “a 1-year ban from arXiv followed by the requirement that subsequent arXiv submissions must first be accepted by a reputable peer-reviewed venue.”2
Accountability, not a blanket AI ban
The policy stops short of prohibiting AI tools outright. Instead, it reiterates that by signing a paper, authors “take full responsibility for all its contents, irrespective of how the contents were generated,” including “plagiarized content, biased content, errors, mistakes, incorrect references, or misleading content” produced by AI.2
Moderation will follow a “one-strike” rule: moderators must document problems and section chairs must confirm them before penalties are imposed, and authors may appeal decisions.2 Supporters see the move as essential to protect research integrity on a widely used, pre–peer review platform; critics worry it may chill legitimate experimentation with AI but acknowledge that blatant, unchecked AI slop is unlikely to remain welcome anywhere.