Arxiv, the open-access repository where researchers publish scientific articles before they have undergone formal peer review, is introducing stricter rules against AI-generated articles containing obvious errors and fabricated content. Researchers who submit texts with clear signs of so-called “AI slop” can now be banned from the platform for a year, according to 404 Media.

Red flags could include, for example, fabricated sources, incorrect citations, or leftover AI comments, Arxiv said. The platform argues that such mistakes indicate the authors have not properly reviewed the AI’s output.

The aim is to counter the growing volume of AI-generated texts that masquerade as serious research.

A single violation could be sufficient for suspension, though a proposed ban can be appealed. Users who have been suspended will also be subject to a future requirement that new submissions to Arxiv must first be accepted by a reputable peer-reviewed scientific publication.