How to Avoid a One-Year Ban from arXiv for AI-Generated Submissions

Introduction

In recent times, AI-generated content—often called “slop”—has infiltrated peer-reviewed literature, causing chaos with fake citations, unedited prompt outputs, and nonsensical diagrams. In response, the preprint server arXiv has announced a strict policy: any inappropriate AI-produced material submitted will lead to a one-year ban from submissions, followed by a permanent requirement that future work undergo peer review before hosting. This guide walks you through the steps to ensure your submissions comply with arXiv’s rules, based on the announcement made by Thomas Dietterich, an arXiv editorial advisory council member and moderator. Follow these steps to protect your academic record and maintain access to the platform.

What You Need

• A clear understanding of arXiv’s submission guidelines (available on their website)
• Access to your manuscript or preprint that you plan to submit
• Knowledge of what constitutes AI-generated slop (e.g., text produced by Large Language Models without human oversight)
• A reliable method to check for fake citations or nonsensical diagrams (e.g., manual review or plagiarism tools)
• An arXiv account (if you don’t have one, you can create it ahead of time)
• Optional: contact information for arXiv moderation team for clarification

Step-by-Step Guide to Avoid the Ban

Step 1: Define Inappropriate AI-Generated Content

Before submission, understand what arXiv considers “inappropriate AI-produced content.” According to the policy, this includes text generated by AI without substantial human editing, fake citations (e.g., references to non-existent papers), unedited responses from prompts, and diagrams that are nonsensical or poorly generated. If your work relies on AI tools, ensure they are used transparently and with significant human correction. If in doubt, review arXiv’s guidelines or contact the moderation team.

Step 2: Verify All Citations and References

AI models often hallucinate citations. Manually check every reference: is the paper real? Does the citation match the content? Use tools like Crossref or Google Scholar to validate DOIs. If you used AI to generate references, replace them with genuine ones. Include only citations you have read or accessed. Fake citations are a major red flag that can trigger a one-year ban and permanent pre-review requirement.

Step 3: Review Diagrams and Figures for Accuracy

Nonsensical diagrams—often produced by AI—are another cause for rejection. Ensure every figure has a clear purpose, correct labeling, and logical data representation. If you used AI to create visuals, double-check that axes, legends, and data points make sense. Human oversight is essential. If possible, have a colleague review your figures before submission.

Step 4: Edit AI-Generated Text Thoroughly

If you used AI assistance to draft your paper (e.g., for grammar or structuring), rewrite each section in your own words. Unedited prompt responses are a key sign of slop. Use a plagiarism detection tool to check for overly generic AI phrasing. arXiv expects human-generated content; even a single paragraph of untouched AI output can lead to penalties. Keep the original ideas but express them with your own voice.

Step 5: Submit Only After Pre-Peer Review (If Applicable)

If you have a history of violations or are unsure about your submission, consider seeking peer review before posting on arXiv. The policy states that anyone caught submitting AI slop will permanently require pre-review for future submissions. To avoid this, voluntarily submit your paper to a peer-reviewed journal or ask a colleague to review it informally. This step is optional for first-time submitters but strongly recommended.

Step 6: Follow Standard arXiv Submission Procedures

When ready, log into arXiv and start a new submission. Follow the platform’s instructions carefully: upload your manuscript as a PDF or LaTeX source, fill in metadata (title, authors, abstract), and select the appropriate category (e.g., physics, mathematics, computer science). Be honest about any AI usage—some fields require disclosure. Avoid rushing; double-check that no AI slop remains.

Step 7: Monitor Your Submission Status

After submission, arXiv’s moderation team may review your paper for compliance. If you receive a message about AI-generated content, respond promptly and provide evidence of human authorship (e.g., version history, edits). Ignoring warnings can lead to an immediate ban. Stay within the community guidelines to keep your account active.

Tips for Long-Term Success

• Disclose AI use: Many journals and preprint servers now require transparency. Mention any AI tools in your acknowledgments.
• Maintain a clean record: A single violation can result in a year-long ban and permanent pre-review. Protect your reputation by submitting only polished, human-crafted work.
• Stay updated: Policies evolve. Check arXiv’s announcements regularly—Dietterich’s thread is just the beginning of stricter enforcement.
• Use AI as a tool, not a crutch: AI can help with grammar or brainstorming, but always edit and verify. The final product must be your own intellectual contribution.
• If banned, appeal responsibly: While not detailed in the original policy, you can contact arXiv leadership to explain your case. Prepare documentation showing your commitment to quality.

By following these steps, you can avoid the one-year vacation from arXiv submissions and continue sharing your research with the global scientific community.