arXiv No Longer Accepts Computer Science Position or Review Papers Due to LLMs (blog.arxiv.org)

arXiv’s computer science category has changed its moderation practice: review/survey articles and position papers will only be accepted for posting after they have completed formal peer review and been accepted at a refereed journal or conference. Authors must include documentation of successful peer review (journal reference and DOI/metadata) at submission; items lacking this will likely be rejected. Conference workshop acceptances do not meet the standard. This is a moderation change—not a new formal policy—and moderators will still accept such pieces only at their discretion, but now rely on outside refereed venues to certify quality. Rejected items can be resubmitted only after an approved appeal once peer review is completed. For the AI/ML community this alters preprint strategy: surveys and position pieces can no longer be broadly circulated on arXiv pre-acceptance, which aims to reduce the recent flood of low-value, often LLM-generated “reviews” and free moderators to prioritize original research. Practically, expect delays in sharing community syntheses and perspectives until after peer review, stronger incentives to publish reviews in established journals/conferences, and potential spillover to other arXiv categories if similar submission surges continue. Authors should plan to obtain and attach DOI/journal metadata when submitting review-style content to arXiv going forward.