ArXiv will ban researchers who upload papers full of AI slop

Attention @arxiv authors: Our policy is that by signing your name as an author of a paper, each author takes full responsibility for the content, regardless of how the content was created.

If artificial intelligence tools produce language that is inappropriate, false, biased, inaccurate, misleading, incorrect, or misleading, and the output is included in a scientific publication, it is the responsibility of the author.

We have recently explained our penalties for this. If the submission has irrefutable evidence that the authors did not look at the results of the LLM generation, this means that we cannot trust anything on the paper.

The penalty is a one-year ban from arXiv followed by a requirement that arXiv submissions must first be accepted in a recognized peer-reviewed journal.

Examples of irrefutable evidence: references, meta comments from LLM (“there is a 200-word abstract; would you like me to change anything?”; “The contents of this table are symbolic, write with actual numbers from your experiments”)