#arXiv imposes one-year ban for unchecked LLM output
arXiv has clarified enforcement for submissions that include unverified large language model (LLM) output. Per a public thread by Thomas G. Dietterich, chair of arXiv’s computer science moderators, a submission that contains “incontrovertible evidence that the authors did not check the results of LLM generation”, examples cited include hallucinated references and leftover LLM meta-comments, can trigger a one-year ban and a requirement that future arXiv submissions first be accepted at a reputable peer-reviewed venue, as reported by The Verge and other outlets. The platform’s Code of Conduct was cited in Dietterich’s thread, noting authors bear responsibility for content irrespective of how it was generated. Community reaction has been mixed, with some researchers supporting the move and others warning about selective enforcement and false-positive risks, according to The Decoder and social reporting aggregated by di.gg.
▻https://letsdatascience.com/news/arxiv-imposes-one-year-ban-for-unchecked-llm-output-be07fdf4
#LLMs #IA #intelligence_artificielle #bannissement #recherche #édition_scientifique #publications #AI #1_an #revues