Preprint Servers Police AI Slop

Preprint platforms are moving from passive repositories to active enforcers of AI provenance by banning or conditioning submissions when papers contain incontrovertible LLM‑generated errors (hallucinated refs, placeholder text, or chatbot meta‑comments). ArXiv's one‑year ban and post‑ban requirement that future submissions be peer‑reviewed is a concrete enforcement model that other repositories may copy or contest. — If preprint servers adopt and enforce AI‑provenance rules, it will reshape researcher incentives, slow unvetted AI‑assisted publishing, and create new governance battles over what counts as acceptable AI use in science.