ArXiv, a leading open preprint repository used by researchers worldwide, announced a policy that could result in a one-year ban for authors who submit papers containing clear evidence that they did not verify text generated by large language models.