ArXiv to Ban Researchers for a Year if They Submit AI Slop
(Friday May 15, 2026 @05:00PM (BeauHD)
from the submitter-beware dept.)
- Reference: 0183225835
- News link: https://slashdot.org/story/26/05/15/1647250/arxiv-to-ban-researchers-for-a-year-if-they-submit-ai-slop
- Source link:
ArXiv says it will [1]ban authors for one year if they submit papers containing AI-generated slop, such as hallucinated citations, placeholder text, or chatbot meta-comments left in the manuscript.
"If generative AI tools generate inappropriate language, plagiarized content, biased content, errors, mistakes, incorrect references, or misleading content, and that output is included in scientific works, it is the responsibility of the author(s)," [2]said Thomas Dietterich, chair of the computer science section of ArXiv, on X. "We have recently clarified our penalties for this. If a submission contains incontrovertible evidence that the authors did not check the results of LLM generation, this means we can't trust anything in the paper." 404 Media reports:
> Examples of incontrovertible evidence, he wrote, include "hallucinated references, meta-comments from the LLM ('here is a 200 word summary; would you like me to make any changes?'; 'the data in this table is illustrative, fill it in with the real numbers from your experiments.'" "The penalty is a 1-year ban from arXiv followed by the requirement that subsequent arXiv submissions must first be accepted at a reputable peer-reviewed venue," Dietterich wrote.
>
> Dietterich told [404 Media] in an email on Friday morning that this is a one-strike rule -- meaning authors caught just once including AI slop in submissions will be banned -- but that decisions will be open to appeal. "I want to emphasize that we only apply this to cases of incontrovertible evidence," he said. "I should also add that our internal process requires first a moderator to document the problem and then for the Section Chair to confirm before imposing the penalty."
[1] https://www.404media.co/new-arxiv-rules-ai-generated-papers-ban/
[2] https://x.com/tdietterich/status/2055000956144935055
"If generative AI tools generate inappropriate language, plagiarized content, biased content, errors, mistakes, incorrect references, or misleading content, and that output is included in scientific works, it is the responsibility of the author(s)," [2]said Thomas Dietterich, chair of the computer science section of ArXiv, on X. "We have recently clarified our penalties for this. If a submission contains incontrovertible evidence that the authors did not check the results of LLM generation, this means we can't trust anything in the paper." 404 Media reports:
> Examples of incontrovertible evidence, he wrote, include "hallucinated references, meta-comments from the LLM ('here is a 200 word summary; would you like me to make any changes?'; 'the data in this table is illustrative, fill it in with the real numbers from your experiments.'" "The penalty is a 1-year ban from arXiv followed by the requirement that subsequent arXiv submissions must first be accepted at a reputable peer-reviewed venue," Dietterich wrote.
>
> Dietterich told [404 Media] in an email on Friday morning that this is a one-strike rule -- meaning authors caught just once including AI slop in submissions will be banned -- but that decisions will be open to appeal. "I want to emphasize that we only apply this to cases of incontrovertible evidence," he said. "I should also add that our internal process requires first a moderator to document the problem and then for the Section Chair to confirm before imposing the penalty."
[1] https://www.404media.co/new-arxiv-rules-ai-generated-papers-ban/
[2] https://x.com/tdietterich/status/2055000956144935055
Too Lenient (Score:2)
by TwistedGreen ( 80055 )
1 year seems lenient considering you can get banned for 7,973 years just for looking at Forza Horizon 6.
Now... (Score:3)
by argStyopa ( 232550 )
...if only our legal system was that stringent?
Ban on practicing law for a year if your submission to the court includes AI slop, how about that?
A second offense, disbarment.
(Personally I think disbarment should be a first-offense result for an ostensibly high-competence field like law, but our society has gotten away from "consequences" for "easily predictable results of ones actions" in general...)
Fix for that (Score:1)
ChatGPT, look at the proposed ArXiv submission and identify anything that looks like "AI slop."
I'm eleventy-ten percent sure someone will try this and twelvity-ten-percent sure it will actually work. /sarcasm
Re: (Score:2, Interesting)
A better idea is to simply have two submission queues. If you get banned from the high quality one you have to use the low quality queue. If they don't do that someone else is going to do it. It is a fact of life that papers are going to have AI generated writing in them going forward. Just accept this fact and make two queues.
Re: (Score:2)
From the summary, it's not AI-generated writing they're punishing but bad output. They're doing it right. It looks like if you generate your paper with AI but review it thoroughly, and get all the mistakes and hallucinations out, it's fine.
Re: (Score:2)
Don't ban from arvix for not proofreading something like citations on a hot result. Where do you draw the line? Drawing the line is hard to do so just push them into the secondary queue. Then they have to redeem themselves to get back into the main one.
This is so obvious, if they start banning people a second competing service will emerge.
Re: (Score:2)
Exactly right.
I really like their approach, and it should be the standard approach for ALL AI generated work.
You can use AI, but you have to verify/validate it. You can't just dump the results without making sure you reviewed the content. Finding obvious stuff that 'would' have been noticed if you actually read through it, and 'should' have been cleaned up is proof that you didn't adequately review it.
What's easier to find, a left-over chat prompt, or the fact that the data was duplicated and all the re
Life sentence for jaywalking (Score:1)
Well, that would be a form of [1]Justice [wikipedia.org].
[1] https://en.wikipedia.org/wiki/Justice_(Star_Trek:_The_Next_Generation)
Re: (Score:3)
> In essence , they're saying that the penalty for sloppy text editing should be (and is) as severe as for, you know, *actual data fabrication*
No. This isn't "sloppy text editing", this is submitting unverified information. Including copy-pasted generative AI content that wasn't verified prior to submission is, by very definition, submitting unverified information. They are saying that the penalty for submitting unverified AI content is as severe as the penalty for submitting any other unverified information. The entire premise is that if you are willing to submit unverified generative AI content, what else is in the paper that wasn't verified?
> What's next, life sentence for jaywalking because "you have shown you're willing to disregard THE LAAAAAW, so we can't trust you not to murder anyone tomorrow"?!
N
Re: (Score:3)
Dismissing fabricated citations as "sloppy text editing" is ridiculous. Citations are effectively you asserting that research evidence exists that backs up your statements and arguments, thus making them little different to fabricating data for a study. Seriously, how is anyone supposed to trust anything in a paper when the literature the study is built off and uses to justify itself are studies which don't exist in the first place? It's a basic expectation of the profession that academics should know the
Re: (Score:2)
But you can submit slop one day a year. Cool. It's like that day once a year where you get to kill everyone that looks at you funny with no consequences.
Re: (Score:2)
I'd prefer if it applied to people who drive below the speed limit, brake at green lights, or who brake going down small hills. That, and those who drive Subarus or Buicks.
Re: (Score:2)
Not sure about the Subarus, but I'll second the Buicks. Also people who hold up traffic to make a right turn from the traffic lane when there's a turning late RIGHT THERE...
Re: (Score:2)
No, they get a 1 year ban, but after that they are permanently barred from submitting anything unless it has been already published in a journal. This reduces by about 90% the utility of the arxiv, and will make them genuinely have trouble getting other people to collaborate when they find out how they are arxiv-restricted.
Re: (Score:2)
Divide by 0?