arXiv Introduces Stricter Guidelines to Combat AI-generated Low-quality Research Papers

ArXiv is enhancing its efforts to address the misuse of large language models in scientific papers.

• ArXiv's Role: Although not peer-reviewed, it is a primary platform for sharing research in fields like computer science and math, influencing what research circulates widely.
• Actions Taken: First-time posters must get endorsements, and ArXiv is becoming an independent nonprofit to manage financial and operational strategies more flexibly.
• New Rules: Thomas Dietterich, head of arXiv's computer science section, announced that submissions including evidence of authors’ failure to verify LLM-generated content could lead to a one-year ban. This includes "hallucinated references" or obvious AI-generated comments, mandating authors to take full responsibility.
• Penalties and Appeals: Offending authors face a year-long ban and must have future papers accepted by a peer-reviewed venue. There is provision for an appeal process following moderator and section chair evaluations.
• Broader Trend: Recent studies have identified a rise in fabricated citations in various fields, often linked to the careless use of LLMs.