The largest preprint platform for computer science, arXiv, has seen a significant surge in paper submissions after the widespread adoption of ChatGPT. To curb the influx of low-quality AI-generated papers, Thomas G. Dietterich, Chair of the arXiv Computer Science Advisory Board, emphasized on social media that arXiv's code of conduct stipulates that once an author is listed as a co-author of a paper, they bear full responsibility for all content, regardless of how it was produced. If generative AI tools produce inappropriate language, plagiarized content, biased content, errors, incorrect citations, or misleading content, and such output is included in the paper, the responsibility lies with the authors. If a submitted preprint contains irrefutable evidence that the authors did not review the output of large models, then no content in the paper can be trusted. For authors found to have such issues, the penalty is a one-year ban from submitting papers to arXiv. After that, to submit a paper to arXiv, they must first have it accepted by a reputable peer-reviewed journal.

https://xcancel.com/tdietterich/status/2055000956144935055