User talk:Gurkubondinn

I'm guessing your username might not mean the cucumber farmer. If you haven't seen this video you might enjoy it. I remember watching Hrafninn flýgur in school in the 80:s, we didn't stop saying tungur knivur and other pseudo-Icelandic for weeks. Happy editing! Gråbergs Gråa Sång (talk) 21:51, 2 October 2025 (UTC)

It actually does mean "the cucumber farmer", but don't ask me why I picked that name 16 years ago because I can't remember why I thought that sounded so clever. :) Thanks for the link, because I remember something from the 90s about prank calls, people named Storm, and "Hej, är det storm?". I've actually only seen Hrafninn flýgur [sv; is] once (but heard the psuedo-Icelandic jokes many times, even in Icelandic) -- with Sólstafir performing a soundtrack live with the movie in the background. There's one decent recording on YouTube, but the one I was was a year or two before. --Gurkubondinn (talk) 11:32, 4 October 2025 (UTC)

		The Patience Barnstar
		thank you for your patience with Gurdeep Hora. his/her replies are clearly AI, and I would not probably have the patience for him/her. The fact that you had to ask nine times is crazy! TornadoElliott (Yelp for help?) 13:52, 23 June 2026 (UTC)

Thanks @Tornado Elliott! This is now at ANI, so I guess it didn't pay off: § Gurdeep Hora. ‑‑gurkubondinn 14:19, 14 July 2026 (UTC)

There is currently a discussion at Wikipedia:Administrators' noticeboard/Incidents regarding an issue with which you may have been involved. The thread is Unsupported and intimidating user-talk warnings during draftification dispute. Ldm1954 (talk) 10:45, 24 June 2026 (UTC)

Saw it, thanks for the notification. Not sure what indicators you're referring to, are you talking about my reply in the WT:AIC § Added page "Semantic ablation" to this project thread? ‑‑gurkubondinn 10:55, 24 June 2026 (UTC)

Yes, you are much better at AI detection than I am. Ldm1954 (talk) 11:11, 24 June 2026 (UTC)

Unrelated, I would be interested in your opinion about this paper on AI. The scanners I routinely use think it is AI. Ldm1954 (talk) 20:46, 25 June 2026 (UTC)

Not having read it very carefully, that comes across as LLM-generated to me. There's some stuff that jumps out at me;

• a bunch of superficial analysis: ,... the model gravitates toward ..., Developers have exacerbated this ..., The logical flow – originally built on complex, non-linear reasoning – ...,
• rules of three: the rare, precise, and complex tokens,
• negative parallelisms (basically the "it's not X, but it's Y" thing): ..., it is not a "bug" but a structural byproduct of ..., ..., they are not seeing improvement; they are witnessing semantic ablation.

Also another (very subjective!) tell is that I start zoning out pretty quickly without taking in any actual information from what I was reading. I think my brain has started to subconsciously recognise this or something.

Did you mean to say "article", or is there a paper somewhere? ‑‑gurkubondinn 14:37, 26 June 2026 (UTC)

That is the paper that Draft:Semantic ablation is based on. An academic in Italy using an AI to write an article on the failings of AI. I do not consider that ethical, but I insist on high ethics in academia. Ldm1954 (talk) 14:43, 26 June 2026 (UTC)

Yeah, thats why I was wondering if there was a paper somewhere (something that looks like a LaTeX typeset PDF with sources), because that looks more like a blog post to me. And I agree with you, academia should be held to very high ethical standards. I didn't spend much time working in academia (sometimes I think that I should have stayed, but then I see stuff like this), and we were certainly held to a higher standard. At least we have decent standard here on Wikipedia, I would have expected universities to also have an interest in this (we're just an assortment of nobodies on the internet). ‑‑gurkubondinn 14:55, 26 June 2026 (UTC)

• Electricity Generation Lanka (edit | talk | history | links | watch | logs)

Hey there. It is not AI generated text. It is text put together by myself, cross checked and verified, only re-worded using a LLM. I believe your action is after this this post? I'd like to kindly request you to reverse this action, or state what exactly is the problem here, when this is 100% human verified and within AI use policy. Thanks Rehman 14:11, 26 June 2026 (UTC)

Yes, that is why I draftified it. As has been explained in that thread, this is not permitted by the WP:NOLLM guideline. Please take a moment to read the guideline, and don't use LLMs on the English Wikipedia going forward. ‑‑gurkubondinn 14:22, 26 June 2026 (UTC)

Fair enough and agreed, and as I said there, AI not my area. That being said, this is a very short article, there are no bloated text or jargon, and everything is human vetted. I have been writing on enwiki for nearly two decades now, you can be assured my intentions are goodfaith. Will you move the page back, or can I? Rehman 14:28, 26 June 2026 (UTC)

For sure, I'm not assuming any bad faith on your behalf. I just assumed that you either didn't know about the guideline or simply had interpreted it incorrectly. It doesn't matter if there is "bloated text" and that everything has been "vetted", the problem in and of itself is just that the text is generated. Someone said (in one of the many onwiki discussions about LLMs on enwiki) that "the standard that you walk past is the standard that you accept", and I've thought about that quite often since then.

Can you just replace the generated text with the text that you had written yourself? Basically the version of it from before you asked the LLM to "improve" it? If you do that, I'd be more than happy to move it back into article space. ‑‑gurkubondinn 14:41, 26 June 2026 (UTC)

Thanks, sure. Let me do that today or maybe tomorrow. Cheers, Rehman 00:36, 28 June 2026 (UTC)

• Draft:Homosexuality in pre-colonial African societies

Can I ask on what basis you determined that this article was 'machine-generated'? I can see nothing obvious to indicate this. AndyTheGrump (talk) 10:17, 28 June 2026 (UTC)

Hey, I just wrote another reply on the WP:LLMN thread as well, but it's not wholesale generated by an LLM. I think that this user is writing some stuff themselves, and asking an LLM to "polish" the text, "format references", or extend it in some other way. I also think that this is a very eager editor (which is good) who genuinely wants to contribute constructively to the project, but guess is that this user believes that using LLMs in a certain way "doesn't count".

Some sections in the article that read more 'LLMish' than others, some sections (and headings) have spelling errors, and some sections have perfect grammar. Some edits switch to using Markdown syntax; Diff/1361151390, Diff/1361218186, and Diff/1361388058 use * instead of '', and other edits like Diff/1359379916 attempt to use '' but use curly single-quotes instead.

There's also somewhat rapid editing going on (though it would be helpful to see the musk article in the contrib history for this), for example the Diff/1359060973 edit added c. 22 kB and was made 35 minutes after the previous edit to the article. Of course it is possible to make other edits in the mean time while you are writing a larger edit (though I would imagine those would be to other articles), and of course there could be other explanations for all of this, but this user has submitted LLM-generated material before.

I don't like accusing people in general, so I have tried to phrase everything as advice rather than as an accusation. They've been given pretty lengthy and good advice by others as well in the past. ‑‑gurkubondinn 10:45, 28 June 2026 (UTC)

As a sidenote, would you mind unlinking the section header (and just put the link to the article at the top of your message or something like that instead)? Check the notes in the blue box on the top of my talk page, I tend to click around with the mouse while I'm reading something and when a section header is linked I will often accidentally forget about that and click it by accident (which is mildly frustrating). ‑‑gurkubondinn 10:47, 28 June 2026 (UTC)

• Sibel Dinçer

The content that's in the article as of now, is everything I've written and added. Previously, I only left the references, external links and categories since I didn't see no LLM detected on there, and for the infobox, I fixed it and changed some things up, but I removed it since the creator added it. So I removed everything that the creator added and kept what I added, and as of now, what's currently in the article is what I added and wrote. MoviesandTelevisionFan (talk) 23:38, 28 June 2026 (UTC)

Anything by this user is presumed to be LLM generated, see the linked AINB filings. Thid is a massive cleanup across hundreds of articles, thousands of edits, and dozens of sockpuppets. When I reinstated the LLMPRV, large portions of the prose were inserted by that editor (have not checked now). You may of course WP:LLMPRVOBJ and take responsibility for it, but I would just like for that to have been done with an accurate rationale. ‑‑gurkubondinn 23:56, 28 June 2026 (UTC)

Hi, noticed your LLM reversion... I take it however the removal of the PROD tag is legit, and so that should be removed again instead of staying restored? Danners430 _{tweaks made} 20:13, 29 June 2026 (UTC)

I've removed all of the ones that were accidentally restored, as far as I know. Except for one, where an article was PRODDED for being LLM-generated from the start, and it doesn't feel valid for another LLM user to remove that. I'm not entirely sure if that's correct though, so feel free to remove it if I am wrong.

Or is there another one that I missed? ‑‑gurkubondinn 20:15, 29 June 2026 (UTC)

Ah, I see what you're saying. I've removed it from George Town Aerodrome, thanks for pointing it out to me. ‑‑gurkubondinn 20:18, 29 June 2026 (UTC)

Please be careful when reverting edits. I 100% support reverting vandalism and LLM usage, but when you revert, you need to make sure there aren't good edits caught in the mix. With a number of your recent reverts, you introduced code that no longer is working resulting in errors. When you revert, please be sure to preview the code to check for Preview warnings and errors. Zackmann (^{Talk to me}/_{What I been doing}) 21:19, 29 June 2026 (UTC)

Aren't you doing those fixes with AWB or something? Then you could easily reapply them. I've reverted hundreds of LLM edits from this editor, and I've fixed dozens of infoboxes after the reverts. I might have missed some, because whenever I can, I use Twinkle to do the reverts so there is no preview to look at. These warnings are not reader-facing, and removing reader-facing LLM junk is more important than warnings that aren't visible to readers.

If you aren't doing this with an automated tool, can you be careful to not fix infoboxes in edits from LLM-wielding users? It makes a it more difficult to fix, especially when the only edits after the LLM users are infobox fixes. If you would for example reference AINB filings, that would prevent this for large cleanups a lot of the time, because it often takes time from the filing until cleanup happens. ‑‑gurkubondinn 21:23, 29 June 2026 (UTC)

And just to check; did you revert any of my cleanups and reinstate LLM text? I don't always get notifications when my edits are reverted, so I need to ask to be sure. ‑‑gurkubondinn 21:24, 29 June 2026 (UTC)

First off, I did NOT revert your edits. No point throwing the baby out with the bathwater. Your removal of LLM text stayed. Just fixed the bad params.

Now some of this can be done with AWB. When I am en-masse making the same change to hundreds of pages, I can certainly do an AWB run. But what is happening is that that AWB run is happening after the LLM use and before you revert. If I do an AWB run on 5,000 pages, I cannot possibly check all 5,000 of them for LLM usage. So when you en-masse revert ALL edits, not just the LLM usage, you are throwing the good out with the bad. When it is a half dozen random pages, there is no point in doing an AWB run as I have to manually figure out what parameters are broken.

Again, I fully support the removal of LLM usage, but you need to make sure that when you revert suspected LLM usage you are ONLY reverting the edits related to the inserting of LLM content. You cannot, for example, revert 10 edits because 2 of them inserted LLM, though it certainly would be much easier...

Again, fully support the removal of LLM and I try to revert it when I find it, though that is not my area of expertise. All I'm asking is that you try to check your reverts to make sure you aren't restoring non-working code. Keep up the good work. Zackmann (^{Talk to me}/_{What I been doing}) 21:31, 29 June 2026 (UTC)

I was just asking because I made hundreds of these edits, and the software doesn't always deliver me notifications when my edits are reverted (despite me having that enabled). I don't know why, but it's not reliable for me for some reason. I did get notifications for the edits where you ping me with my username (that has previously often not worked for me either, but apparently it does now!)

Like I said, I try my best to not reinstate old parameter names. But sometimes it happens, especially when I can quickly revert something with Twinkle. Checking the preview each time would make it take much longer, and my brain already feels like total mush after slogging through just ~15% of this user's edits.

When it's LLMPRV edits, it is more than just "suspected" edits btw. Those are users that we know have been adding LLM to enwiki, and this process exists to make it realistic to get it out of the encyclopaedia. Some good edits will unfortunately sometimes be casualties of this process, but that's just the reality of trying to deal with this literal flood of junk. I have one list of 600 articles from one user (that was autopatrolled!) to go through. I just found one another autopatrolled user with 150k edits pasting raw chatbot output into Wikipedia, and there is an open AINB filing for an admin.

But the ones that you have fixed with automated tools can just be re-fixed again. At least for the most part, sometimes there might be different parameters pre-LLM user and post-LLM user. At least none of that is reader-facing, which is the important part. If I'd see an infobox being broken for the reader after reverting, I would of course fix that immediately. When I see your edits in the history list, I try to check and fix as I can. I fixed a couple of them by hand today during the reverts, you can Ctrl+F my contribs for "infobox" and that will show you some of them. When I'm aware, I fix it/don't break it. I'm not reintroducing deprecated parameter on purpose, that I can promise you. I am slogging through LLM edits, and it feels like each one of them makes me dumber than the last one.

You cannot, for example, revert 10 edits because 2 of them inserted LLM, though it certainly would be much easier...

When you see edits like this, it is because the edits between the LLM-user edits only edited text that the LLM-user had added. But then I've read through the diffs in between to check for that. I hate doing those pages, because they take a much higher mental load, and this is daunting enough as it is.

And btw, I'm in favour of cleaning up deprecated parameter. I'm a programmer and, unlike all of the vibecoders out there, I think that maintainability matters. Thanks for doing this, and know that I am not trying to sabotage you. Sometimes we'll just make mistakes that create a bit of extra work for each other, and I'll do my part to try not to create extra work for you to the extent that I can. ‑‑gurkubondinn 21:45, 29 June 2026 (UTC)

Bottom line, we are all on the same team here. The thing you are missing is that the old AWB run cannot always be reused. There are hundreds, if not thousands, of custom AWB runs that I have cooked up. So when a half dozen pages with one Infobox type come up, I would have to either find that specific custom AWB run's code, recreate it, or just fix it manually. There is not magic button to push that fixes ALL deprecated parameters on ALL pages. Much more complicated than that.

In any case, all I'm asking is that you try to check for these parameters, which it sounds like you're going to do. That is all I can ask for. Keep up the great work! Zackmann (^{Talk to me}/_{What I been doing}) 21:57, 29 June 2026 (UTC)

Quick follow up, in addition to the above, I want to be clear... NEVER thought you were intentionally introducing bad params. I totally get it was a side effect of your good work reverting LLM. Zackmann (^{Talk to me}/_{What I been doing}) 21:58, 29 June 2026 (UTC)

I appreciate you!!! Zackmann (^{Talk to me}/_{What I been doing}) 18:10, 7 July 2026 (UTC)

Hello Gurkubondinn,

Thank you for reviewing the Geoffrey Budworth and Des Pawson articles.

I first researched these two authors and created the articles on the French Wikipedia. I then used AI to translate them into English, as English is not my native language. Before publishing them, I proofread, corrected, and supplemented the English versions.

My intention was not to publish automatically generated articles, but to make articles I had written in French, based on reliable sources, available in English.

I understand that the English versions may not yet meet the editorial standards of the English Wikipedia. My goal is to improve them, not to republish them as is.

Could you please tell me the main points that you think should be corrected before requesting their return to the main Wikipedia page?

Thank you in advance for your time and advice.

Nef (talk) 00:18, 30 June 2026 (UTC)

What I think should be "corrected" is that they should not be generated, like they are now. If you lack the WP:COMPETENCY to contribute in English using your own words, then you should not be editing the English Wikipedia. ‑‑gurkubondinn 00:29, 30 June 2026 (UTC)

I understand your position. My goal was simply to make articles that I wrote in French available in English.

I will consider the best way to proceed in order to meet the expectations of the English Wikipedia.

Thank you for your feedback.

Nef (talk) 00:40, 30 June 2026 (UTC)

Draft:Harsh Roshan

Harsh Roshan is indian famous actor...There are many articles about him on well known, trustworthy websites. Anyone can check them out via Google and improve this article, right? Botempa (talk) 09:33, 30 June 2026 (UTC)

The reason that I declined that draft is that it fails WP:NOLLM and the WP:V policy. It was not about the subject of the article (the actor), it was just about the article itself and how it was created. ‑‑gurkubondinn 09:39, 30 June 2026 (UTC)

Let me be honest about this. I enjoy creating articles... I search for famous people on Google and look for reliable sources of information about them. If I find such sources, I create an article about them. Since the articles I find are usually in English, I translate them into Sinhala.

Because as I said before, my mother tongue is not English... Sinhala is my mother tongue Because I know very well that AI is not 100% successful... That's why I translate them.. If you don't know how to do it, get help through AI... What I'm saying is, check this article and check the sources.. The information in the sources is the same. The article also I would like to say honestly... Even though I don't have much experience..

Please you check reference and check article carefully..and this article other editors fix Botempa (talk) 09:53, 30 June 2026 (UTC)

If you don't know how to do it, get help through AI...

No, don't do this. You need to actually read our documentation and understand it. The chatbots frequently give people bad or flat-out wrong advice about how Wikipedia works. Sometimes they even make up stuff, and then these users come here and quote non-existent guidelines or claim that policies say something that they don't. You need to read this yourself, there are no shortcuts here, so you will also need to be able to understand them.

Like I told you at ANI, you won't actually learn English if you keep leaning on LLMs instead. I am myself learning a language right now. I go to classes four days per week. It takes a lot of effort, I am not a natural at learning languages, and it takes a lot of effort for me. That is okay. But you (and I) need to do the work ourselves. Otherwise we learn nothing.

What I'm saying is, check this article and check the sources.. The information in the sources is the same.

I did, and the first source that I checked did not support what was sourced to it. The draft claims that Cinema Express noted that he delivered an earnest performance, ..., but in fact the source used earnest about someone else's performance.

‑‑gurkubondinn 10:00, 30 June 2026 (UTC)

I won't create articles until I get good at English..I asked if it's possible to fix the shortcomings of this article. Because there are reliable sources..

Then we'll meet again one day... I'm going to Sinhala Wikipedia because it's my mother tongue..

You all have more experience than I do, so please forgive me for any mistakes I have made.

Please do not continue this from today onwards. I may be at fault, but it hurts when different people criticize me. I only did this because I genuinely enjoy it. I am sorry.

I will never forget your feedback. I will take it to heart and use it to improve in the future..... Botempa (talk) 10:11, 30 June 2026 (UTC)

Of course it's possible :) You (or someone else) just have to find the sources needed and summarise them. It might even be possible with the sources that are already present in the draft, but I didn't check the rest of them.

You should also read the WP:BACKWARDS essay. You should first find the sources and read them, and then summarise them. Not start by writing what you want to have in the article and then look for sources to support that. Start with the sources.

I am glad that you enjoy it. I am also here because I enjoy Wikipedia. I want to help you, because I can clearly tell that you are here because you want to be here, and because you enjoy doing this. That's why I'm showing you all of this stuff. It is to teach you, not to criticise you. There are a lot of policies, guidelines, and documentation here, and it can be overwhelming to understand at first. Nobody knows all of this stuff when they start editing, and I didn't. You learn it gradually. You just shouldn't be asking AI for help with it, because it is genuinely bad at it. People managed before the widespread availability of these chatbots, and those that have been relying on them have consistently shown that they do a very bad job. So it is better to do it like folks always have done it here.

And I'm glad that you are going to be contributing to the Sinhala Wikipedia. They need editors. I sometimes contribute to the Icelandic Wikipedia (and Swedish Wikipedia), which also needs editors. But I've found that I enjoy the larger English Wikipedia more, so this is where I spend most of my wikitime.

We only have this experience because we have spent time on building it. There is nothing special to it. And all of us have made mistakes (I once made someone very angry by moving Hitler Masturbating to draft space because I thought it was not real, you can read it in my talk page archives). That happens to everyone. Nobody is angry at you, and I certainly am not angry at you. There's no need to ask for forgiveness, but if you want it, then you have my forgiveness. But I don't think that you did anything terrible, you just were not aware of this stuff and made honest mistakes. What matters is that you learn from them, listen to us, and don't go and repeat those mistakes. It's only a problem when people ignore the advice that they have been given and continue to do the same stuff again and again, because then it is not a mistake.

‑‑gurkubondinn 10:20, 30 June 2026 (UTC)

Thank you for understanding me, my friend.... I used to spend most of my time on the English Wikipedia, but now I will visit occasionally to fix spelling errors and make corrections..

There is no one active on the Sinhala Wikipedia, just as you said.. I will constantly help with that.

Let's meet again someday, my friend.

Thank you to everyone for pointing out my shortcomings, I am leaving happily.. Botempa (talk) 10:30, 30 June 2026 (UTC)

Hi. Thanks for flagging that Riverdance AI content. I wanted to get your thoughts on this user and whether you thought it may be in the same boat or same person as the first user. The edit summaries and content amouunts are very similar and the main topic and article was similar. Could this also be LLM edits? Should more content be reverted at Anúna? Thanks. DaHuzyBru (talk) 12:08, 30 June 2026 (UTC)

All I can say at a glance is that it seems possible. But I don't have time right now, but I'll take a closer look later. You can also post on WP:AINB if you want to get more eyes on it, there are some very capable people there. ‑‑gurkubondinn 12:13, 30 June 2026 (UTC)

Hello there!

You reverted an edit on Eurodicautom and warned the editor on their talk page. Said editor (if you noticed) had also been warned months prior about LLM usage in their article creations. I was noted of them from an edit they made to Renaissance in Ferrara. Also "cleaning up" AI text. I'm not one to assume but what's your thoughts on bringing the issue to WP:AIN? Love seeing the AI maintenance templates gone but a "fixing" using AI is still using AI, yeah?

PeepeeDino (talk) 14:20, 2 July 2026 (UTC)

Yes, I think that would be appropriate. They seem to have been removing the {{AI-generated}} tags while using LLMs themselves, so this user's history definitely need some looking into. ‑‑gurkubondinn 16:13, 2 July 2026 (UTC)

Very well then, I'll draft something later in the day. Would you mind if I pinged you in the discussion for any further evidence of their AI usage? If you encountered them in any other instances that is. I'm not too good at spotting the more subtle signs of LLM writing apart from the markdowns and flowery prose tbh.

PeepeeDino (talk) 22:59, 2 July 2026 (UTC)

Sure, feel free to ping me. I spot a lot of it using those signs too. I already spend far too much time on this, trying to find the subtler signs would take up more time there there is in a day anyway. I just notice it when I start looking into a particular user and looking at their contributions and checking for source integrity issues and the like. ‑‑gurkubondinn 23:04, 2 July 2026 (UTC)

I'm in WP:3RR territory but wanted to make sure you saw Special:Diff/1362310167/1362369833... Zackmann (^{Talk to me}/_{What I been doing}) 15:42, 3 July 2026 (UTC)

Ah, thanks for reminding me about that one. I gave up after two reverts. Was going to give them a chance to reply in User talk:Politicalenthusiast556 § July 2026, but they've ignored me and continued edit warring to keep their AI-generated version. ‑‑gurkubondinn 15:47, 3 July 2026 (UTC)

Since they have ignore my question, I'll bring this up at AN3 unless you are against that? I've always understood 3RR to be ${\displaystyle RR\geq 3}$, but as a rule I stop after two reverts because at that point it just isn't productive.

I saw that you had RfPP the Shandong cliqueShandong clique redirect where something similar was going on, thanks for taking care of that. ‑‑gurkubondinn 16:00, 3 July 2026 (UTC)

I just don't have the energy for this fight right now... Lol. Figured I'd let you handle it! I endorse whatever action you take. Zackmann (^{Talk to me}/_{What I been doing}) 16:03, 3 July 2026 (UTC)

I've nudged them again on their talk page, and I'll notify AN3 if they continue ignoring me. ‑‑gurkubondinn 16:12, 3 July 2026 (UTC)

Went to RfPP rather than AN3, since they haven't edited since their most recent revert: Diff/1362390697. ‑‑gurkubondinn 17:24, 3 July 2026 (UTC)

that's generally interpreted as a requested WP:G7, and there's no need to restore AFC templates. The point of the "do not remove this section" comment is to prevent editors from making it look like a draft hasn't ever been declined, which isn't the issue if they blank the whole page. If you stumble on something like Draft:PrivacyEngine again the most helpful thing to do here is to just add the CSD tag. Rusalkii (talk) 21:42, 3 July 2026 (UTC)

If I recall correctly, the reason I did that was to make it more easily identifiable that some user had been using LLMs and it was somehow unquestionably identifiable in that draft. But honestly I don't remember, normally I would G7 it or just let it be until G13 kicks in. ‑‑gurkubondinn 21:54, 3 July 2026 (UTC)

Were you pinged from my comment? I didn't get a notification. I added {{reply to}} by edit so not sure if it worked. Dw31415 (talk) 14:29, 4 July 2026 (UTC)

No, that never seems to work for me. You're supposed to get a ping if someone uses [[User:Example]] in an edit summary, but it doesn't seem to work for me for some reason. ‑‑gurkubondinn 14:32, 4 July 2026 (UTC)

Ah, I see what you're saying. You're talking about the pings in Diff/1362520036? Those won't work, because you didn't add a signature in the same edit. You can do a WP:PINGFIX to send out the pings. ‑‑gurkubondinn 14:33, 4 July 2026 (UTC)

Thanks! I moved them to a new comment and got confirmation. Dw31415 (talk) 14:47, 4 July 2026 (UTC)

I'm not sure what you meant by:

not sure what you wanted to do with {{efn}}, but it's been replicated a couple of times in the {{noteslist-talk}} below (doesn't show inside the {{textdiff}} without using something like nowiki tags

I'd like to keep the diff clean and simple to hopefully avoid objections or a longer process. Dw31415 (talk) 14:48, 4 July 2026 (UTC)

I just mean that the {{efn|...}} isn't visible in the {{textdiff}} itself, it gets rendered by the parser and as an actual footnote instead.

{{textdiff|1=Text generated by large language models (LLMs)<nowiki>{{efn|such as ChatGPT, ...}}</nowiki> often violates ...|2=Text generated by ...}}

You would have to do something like that with a <nowiki> tags if you wanted the {{efn|...}} to be visible. ‑‑gurkubondinn 15:08, 4 July 2026 (UTC)

You're right about the bot misfiring. If you want to report it, it looks like you'll have to post a message and email User talk:Σ. Wild guess at the bug: the pipe in <code>|hat=no</code>? Boud (talk) 22:40, 5 July 2026 (UTC)

My guess is that it's a timing issue or race condition. You added your reply right before it fired, and it may read pages in batches and then archive/remove a few seconds later or something, causing it to have also removed your reply. But who knows, might be worth reporting this error, but it's too late for here me to want to deal with it right now :) ‑‑gurkubondinn 22:46, 5 July 2026 (UTC)

hi sorry but that is just how i write. look at other articles i have written such as XCOMM, Evil Island or Swollen Teeth. i apologise for sounding robotic when it comes to articles but yeah if i keep rewriting this over and over there would be more accusations of me using llm for wikipedia articles.. i can't really think of other words to even replace the originals in the sentence "the band gained further recognition", as it was literally recommended by my iphone's keyboard! yeah i do use ai to help me get sources (i check those) but i do NOT use ai to write articles. i have acknowledged that rule since it was there. Dehydrothiotoluidine (talk) 11:05, 6 July 2026 (UTC)

It's not about you "sounding robotic", it is about edits like Diff/1362471369. ‑‑gurkubondinn 11:41, 6 July 2026 (UTC)

oh yea, ya, that was me using ai as an example. i apologise and i ran out of ideas that time. but yeah i didn't use ai "wholely" to write all words in the article i just "referenced" that text and rewrote it. forgot to add that while writing the message above yoursDehydrothiotoluidine (talk) 11:45, 6 July 2026 (UTC)

So in other words, you used an LLM. This is prohibited by the WP:NOLLM guideline here on the English Wikipedia, and that is why I declined your draft. Please just abandon any and all use of these tools if you want to contribute to the project. ‑‑gurkubondinn 12:06, 6 July 2026 (UTC)

alright, thanks for the advice and reply! i think i'll try not to, to prevent things like this from happening ever again... well getting a draft declined or an article put into drafts, is disappointing Dehydrothiotoluidine (talk) 12:17, 6 July 2026 (UTC)

note: i took example from ai instead of fully incorporating the text it generated into the article Dehydrothiotoluidine (talk) 12:38, 6 July 2026 (UTC)

And what does NOLLM say about "taking examples" without "fully incorporating" it into the text? Are there any exceptions to "incorporate" text in any way, or "taking examples"? Or is it simply prohibited? ‑‑gurkubondinn 13:51, 6 July 2026 (UTC)

"Editors are permitted to use LLMs to suggest basic copyedits to their own writing, and to incorporate some of them after human review, provided the LLM does not introduce content of its own." — (Writing articles with large language models) Dehydrothiotoluidine (talk) 04:05, 7 July 2026 (UTC)

That means that you are allowed to use it to suggest those things to you, which you then need to implement yourself. It does not allow you to copy LLM output like you did. The answer to the question that I asked you is that it is prohibited. ‑‑gurkubondinn 09:06, 7 July 2026 (UTC)

• Wikipedia:Articles for deletion/Francesco Gargallo di Castel Lentini

Are you sure it's best to curtail the AFD instead of letting it run? The page absolutely fails notability and I believe it would be better to have it deleted right away. The AFD can be reopened and the page un-draftified. Geschichte (talk) 06:44, 7 July 2026 (UTC)

Hi @Geschichte, that was a mistake. Should I move it back, can the AfD be resumed? ‑‑gurkubondinn 09:04, 7 July 2026 (UTC)

Yes, the AFD can continue if the page is moved back to mainspace. In this case the close of the AFD can be reverted. Otherwise, many thanks for your attention to the creator of this page. Geschichte (talk) 11:42, 7 July 2026 (UTC)

In that case I'll revert it, I made a mistake in draftifying it with an open AfD anyway. And deleting via AfD is better than draftification too. ‑‑gurkubondinn 11:51, 7 July 2026 (UTC)

 Done: Diff/1362988856

‑‑gurkubondinn 11:55, 7 July 2026 (UTC)

No objections on my end. — Godsy ^(TALK_CONT) 12:00, 7 July 2026 (UTC)

Great, thanks for taking care of it. Apologies for messing this up. ‑‑gurkubondinn 12:26, 7 July 2026 (UTC)

Thanks, and interesting to note the behaviour on no:wiki. He tried to add a huge chunk of material ten times in one night, the most surprising thing is that someone was awake to revert as the clock was nearing 03 in the night. Geschichte (talk) 20:48, 7 July 2026 (UTC)

You are blaming me of misusing LLM, with this unsibmitted example. However, when submitting the Drievliet draft, it was all done according to Wikipedia:LLM-assisted translation. ~2026-35805-86 (talk) 10:00, 7 July 2026 (UTC)

Yes, because you have. There are very strict requirements for LLMT, not all of which you have followed. There are also additional AISIGNS in many of your edits, including (but not limited to) the incontrovertible sign in the Drievliet draft that you linked to. Please don't use LLMs on the English Wikipedia, it is prohibited by the WP:NOLLM guideline (and yes, I am aware of LLMT, which is how I mentioned it), and I am worried that you may get blocked if you continue. ‑‑gurkubondinn 10:06, 7 July 2026 (UTC)

Yes, and that is also the reason I started writing you. As far as I know it's allowed to use LLM for translating articles, using Wikipedia:LLM-assisted translation. Note that when I submitted the draft I don't see issues of something I did wront. I also added additional sources compared to the Dutch page for better referencing. I also didn't translated the whole page, and didn't include non-references parts. The issues you raised when reviewing were two non functioning links, that I left in the article per WP:Translate Include the same sources that the original article used, if they are appropriate.. ~2026-35805-86 (talk) 10:16, 7 July 2026 (UTC)

My personal opinion is that no translation is better than a machine translation (especially when done with an LLM). If someone wants to read a foreign language article, then they could just put it through a machine translation tool themselves, there's no benefit to someone doing that and saving it on enwiki. Note that this is my opinion, not a guideline or policy.

Translating a body of work is not an easy, simple or quick task. Other Wikipedias have their own policies and guidelines that may differ from the enwiki guidelines and policies in various ways, so content from there doesn't automatically conform with the enwiki policies and guidelines. I don't speak Dutch or French so I am not all familiar with the policies and guidelines on frwiki or nlwiki. They might very well be pretty similiar, I just don't know because I'm not active there.

Include the same sources that the original article used, if they are appropriate.
— WP:TRANSLATE

I understand your good intentions here of following policies and guidelines, but WP:V is a core policy and takes precedence. That applies for sources that work and are used correctly (in line with enwiki's policies and guidelines), which is probably why it is hedged with the if they are appropriate bit. You can of course use references from the translated article, but if they are dead then you need to use archived-url to include archived copies of them instead so that you are citing verifiable sources.

But I understand where you are coming from. Think of it this way; translating something needs to also follow all of the same policies and guidelines that writing something yourself needs to follow.

3. You have checked the sources in the origin language article and you are sure the translated text reflects them fairly, ...
— WP:LLMT

When you include references that return 404 not found, WP:DEADREFS, or don't exist, then this requirement has clearly not been met. You can't include unverifiable references just because they are in the article on nlwiki or frwiki.

You should also get into the habit to WP:ALWAYSARCHIVE any sources that you cite, also ones that are live. Sources go offline all the time, websites change their URL structures and don't bother forwarding from the old URLs. The content and text on sources can also change over time, so including a link to an archived snapshot of what source looked like and what it said when you sourced a claim to it helps with WP:VERIFIABILITY as well.

4. You have complied with all the usual translation processes including terms of use-compliant attribution.
— LLMT

This was done on the Drievliet, but there were other drafts/edits where you only said this after I had reverted or tagged. This is actually more important than it might seem to, because it is a legal copyright requirement. As far as I understand, it is legally possible to do this afterwards, but you should try to always do it at the time that you add translated text. Copyright is complicated.

‑‑gurkubondinn 10:47, 7 July 2026 (UTC)

Same or similar situation, but a different article: Diff/1362986225. Did you read the report that you are citing? I see that the original URL for the report is dead and that there's an archived url, but there's one clue to me that indicates that you may not have read it.

And if I may ask, do you speak Italian, French, and Dutch in addition to English? I speak five languages (three at a native level, and actively learning the fifth one), so I'm not saying that it's impossible, but I wanted to check. ‑‑gurkubondinn 11:24, 7 July 2026 (UTC)

This was actually problematic enough that I've fixed it: Diff/1362987034. This is one of the reasons that machine translations are bad, and why you are required to read the sources yourself. It's no different from writing something yourself, you may not cite any sources that you have not actually read yourself. ‑‑gurkubondinn 11:28, 7 July 2026 (UTC)

Yes I checked the source, you can read it here. I checked the content, but not every detail. I speak Dutch, French, German and English and a bit Afrikaans. I'm not native English, so I don't recognise your example as "Will you be able to land from that position?". I never change the access-date when translating, as it was written from the version at that date. ~2026-35805-86 (talk) 12:28, 7 July 2026 (UTC)

The access dates indicates when the source was read and verified to match the article contents. Please read the Template:Cite web documentation for example.

And I did check the report, that's how I know that your edit misrepresented the direct quote. The FO's question is also a direct quote in the report:

The First Officer asked the Captain "Can you make it to land from that position?"

What I think happened is that you gave the itwiki text to an LLM and requested a translation, and it translated the (already translated) direct quote from Italian back to English. So the sentence Can you make it to land from that position? had first been translated to Italian, then it was machine translated back to English and became "Will you be able to land from that position?". If you had checked the source, then you should have seen this. ‑‑gurkubondinn 13:26, 7 July 2026 (UTC)

Yeah, I know that the repeated readdition of categories to a draft is suggestive of LLM use, but that doesn't really change too much since I still shouldn't have to patiently put up with pulling it out of categories over and over again without escalating to warnings. And even though the IA's were probably related, and thus probably the same person, the use of different IA's each time they edit suggests that they wouldn't see the note if I had gone the traditional route of warning them on their personal talk page, because they wouldn't be on the same IA again the next time, which is why I opted to post it as a comment directly on the draft. Bearcat (talk) 14:01, 8 July 2026 (UTC)

Yep, and agreed with everything. I've just seen you getting increasingly annoyed at this lately, and just wanted to make sure you're aware of why this always seem to be happening increasingly often. ‑‑gurkubondinn 14:28, 8 July 2026 (UTC)

Who was this addressed to? Selbstporträt (talk) 14:08, 9 July 2026 (UTC)

The complaint of identify the section in Diff/1362670756, that was addressed in Diff/1362712625 and on Talk:Prelude to the 2026 Iran war § LLM hat. You removed the reason parameter (along with the tag) in Diff/1362786940, and didn't include the reason parameter when you tagged the sections in Diff/1362786940 and Diff/1362789087. You also omitted the date parameter, causing AnomieBOT to set it to {{CURRENTMONTHNAME}} {{CURRENTYEAR}} in Diff/1362790928, instead of March 2026. The affects which subcategory of the Category:Articles containing suspected AI-generated texts maintenance category the article is included in. Why do you ask?

Btw, the documentation is a bit unclear (as you pointed out), but to tag a section you should do

{{AI-generated|section|date={{CURRENTMONTHNAME}} {{CURRENTYEAR}}|reason=The AISIGNS, diffs, etc}}

The first unnamed parameter is a string that (defaults to article) that affects the This article may contain ... wording in the article. If you set it to section it becomes This section may contain .... But you can set it to whatever you want, test page becomes This test page may contain ... and so on.

There is also a section parameter which will automatically set the tag text to This section may contain ..., but I haven't worked out if it's supposed to be a yes‍/‍no boolean, or if it is supposed to be the section name, or something else. So I never use it. ‑‑gurkubondinn 14:25, 9 July 2026 (UTC)

I asked because I have no idea who you were addressing. Since I'm the one editing the page, it could have been me. But then I don't recall never complaining about any parameter, let alone an empty one. Why would I? I'm not the gnome here. You are, and your response shows me you do care about parameters!

My own issue is with drive-by patrolling. When I tag a page, I read it; and in fact when I do tag a page, I try to resolve the issue. I know that's not gnomes' thing. Yet we can't hide that lack of due diligence with a "may". Even if I could bypass the difficulty I was dealing with (excerpts) using a nifty trick, the issue remains. Tagging a whole page because one editor has been accused of using LLMs should not be the best practice from our LLM patrol.

Look. I like gnomes, and I like your page. I'm sorry if my lack of technical savvy messes with your work. I'll do my best not to forget, and will immediately add what you just said in my editing kit. To give you an idea, I just spent 15 mins on knowing how to link to the Diff.

In return, I'd like you to make sure that your hat game be as precise as possible. That is, if a section is LLM generated, then identify the sections. So at the very least, if you want to know who wrote what, then use Who Wrote That? Simpler than Blame, which would be overkill.

Deal? Selbstporträt (talk) 14:51, 9 July 2026 (UTC)

That's fine, the parameter stuff didn't bother me so much. It was more about the combative tone, combined with removing what you had complained about. I just explained the parameters to you here for future reference, because you had said in one of the edit summaries that you weren't sure how it worked.

But I can assure you that this wasn't a case of "drive-by tagging" (and even if it were, you would be barking up the wrong tree here because the tag wasn't even added by me). Look into the history of the user that created the article and the sockpuppet investigation. I don't know much about this case (I think it's pretty much only CUs that know the full details), but it's definitely not great. Personally, I think that the whole article should be tagged because of this, so that the readers are aware from the start. Or that the whole thing should be deleted, or the sections left from these sockpuppets completely removed, which can be done under LLMPRV.

Also keep in mind that the main purpose of these tags are reader disclosure (and maintenance), to let our readers know that they may be about to read generated text, and to let them know about some of the more common problems with such text. ‑‑gurkubondinn 15:24, 9 July 2026 (UTC)

If you continue to hold that "I don't know much about the case" and "I think the whole article should be tagged", then we indeed have a problem. At least one bigger than tone, for I think this makes no pragmatic sense at all, and indeed constitutes drive-by tagging. Either you read the page, or you don't. If you don't, then you're just driving by.

The presumption of removal applies to what the suspected editor did, not other editors. That's important here because that page is a child of another, as you already know. Stuff from 2026 Iran war has been split there. If you delete that page, then you're failing to preserve. This is not great.

The main issues with LMMs are hallucinations and sycophancy. The first one is easy to check. The second one requires some copy-editing. Either way text needs to be read.

All the Iran pages are not in great shape. We can't blame LLMs for that! Selbstporträt (talk) 17:55, 9 July 2026 (UTC)

Again, it wasn't me that tagged the article. And look into the user account, you have made some incorrect assumptions here. This isn't your average LLM users, or your average normal users for that matter. If I remember correctly, this happened shortly before the war started. ‑‑gurkubondinn 17:59, 9 July 2026 (UTC)

So you still haven't looked at any edit. Fine. Then let's hope we don't meet again.

Good bye. Selbstporträt (talk) 13:45, 10 July 2026 (UTC)

Yes I have. I can guarantee you that this user (across multiple accounts) was using an LLM. This is not a debatable opinion. But if you don't want to look into this user then I can't help you. Please don't disrupt AINB cleanup or remove tags like this in the future. ‑‑gurkubondinn 13:47, 10 July 2026 (UTC)

For one last time, "was using an LLM" ain't enough. It's "was using an LLM on the edits of the page", with specific diffs.

Not only I will delete a misplaced and unjustified hat regarding LLMs, next time I will report anyone who misrepresents LLMPRV to threaten to delete edits made by uninvolved editors. Selbstporträt (talk) 14:07, 10 July 2026 (UTC)

No it isn't. Just because I can't detail a SPI case, when I am not a checkuser, because I don't remember what was public and what I figured out, doesn't mean that somehow LLMPRV is invalid. This isn't "misrepresenting LLMPRV", nobody is "threatening to delete edits made by uninvolved editors". I've answered your questions, given you help, looked up edits for you, and you are still combative and difficult to deal with. Please stop now. ‑‑gurkubondinn 14:12, 10 July 2026 (UTC)

If you had bothered to look up the SPI case and figure out something about this group of sockpuppets, you might actually have been less difficult. But please stop now. ‑‑gurkubondinn 14:13, 10 July 2026 (UTC)

The main condition for LLMPRV is being blocked for using LLMs. We don't even have that.

The "is the only significant contributor" clause should be read very conservatively, not liberally. Selbstporträt (talk) 14:25, 10 July 2026 (UTC)

No, that's one of the conditions. There are other conditions listed in WP:LLMPRV § Requirements (1A through 1D), for example consensus can be established or the editor has been inactive for a certain amount of time. There no exceptions for "blocked for other reasons" or something like that. The whole purpose of LLMPRV is to be a tool to deal with the asymmetry of effort that is inherent to LLM generated article and edits.

And nobody has said "this article should be deleted under LLMPRV". I expressed a personal opinion that this article should probably be deleted, and told you that this personal opinion is based on the article having been generated by an LLM and inserted by an account that was a sockpuppet. If I actually wanted to have the article deleted under LLMPRV, then I would have tagged it with {{prod llm}}. But I haven't because I haven't argued that it should. ‑‑gurkubondinn 14:38, 10 July 2026 (UTC)

Hi @Selbstporträt, jumping in here as a CU quite familiar with this case, and I can assure you that LLMPRV is absolutely valid for anything produced by this cluster. They have been extensively using LLMs to write pov-pushing content across multiple languages. They weren't blocked specifically for using LLMs, that's true, but that's because they were WMF-banned before ChatGPT was even released. In solidarity, asilvering (talk) 14:42, 10 July 2026 (UTC)

"that's one of the conditions"

Which part of "The following requirements must apply" you do not get? Selbstporträt (talk) 14:48, 10 July 2026 (UTC)

Read further, requirement 1 is:

1. The editor who added the content either:[emphasis added]

There are four distinct requirements (1A through 1D) concerning the state of the account. The operative word here is either, meaning that any of the criteria 1A through 1D need to be met.

Requirement 2 is about users with a recent blocks that are currently active. Requirement 3 is basically what enables you to deal with their edits as a whole without needing to justify it on a per-article basis, to deal with the asymmetry of effort. ‑‑gurkubondinn 15:04, 10 July 2026 (UTC)

Good. Which of the criteria through 1A through 1D has been met: someone who's inactive because he's been blocked for other things, perhaps? That'd be just great. How about "discloses that they have used AI and consents to clean-up of their past edits": why in sweet hell should testimony to clean up one's act be taken as evidence for complete erasure - it was in fact a struggle session? The last clause is obviously self-fulfilling: it applies performatively, or eo ipso. Irrelevant here.

So none of the requirements should apply.

Requirement 3 is more interesting. It is specifically about the contributions to be removed. If you say that the whole page needs to be tagged, then you interpret "is the only significant contributor" in the loosest sense possible. The page as a whole could be LLM-infected because presumably because it's been authored by someone who uses LLMs.

That should never, ever apply to a page that has been edited by 79 editors and has become a split from an overloaded main page.

Before you start counting characters, bear in mind that I mostly deleted stuff. That work counts too! Selbstporträt (talk) 15:52, 10 July 2026 (UTC)

"I haven't because I haven't argued that it should"

You only argued that the tag should apply to the whole page! Selbstporträt (talk) 14:51, 10 July 2026 (UTC)

Yes. Exactly. I even told you why. ‑‑gurkubondinn 15:05, 10 July 2026 (UTC)

So now you're going to resist even the most basic inference.

Then let's quote that reason:

I don't know much about this case (I think it's pretty much only CUs that know the full details), but it's definitely not great. Personally, I think that the whole article should be tagged because of this, so that the readers are aware from the start.

So the reason is that you're interpreting "is the only significant contributor" in the loosest sense possible. Just like I said.

You realize other patrolling knights could ask for the whole page to be deleted because of that tag, right? Selbstporträt (talk) 15:57, 10 July 2026 (UTC)

I'm sorry that I ever tried to explain anything to you. I wont try that again. Please leave me alone now. ‑‑gurkubondinn 15:59, 10 July 2026 (UTC)

why did you take away incorruptibility of Meivazhi? Wisdomsalai (talk) 17:13, 12 July 2026 (UTC)

Hi there @Wisdomsalai, that is explained in User talk:Wisdomsalai § July 2026. Using AI-tools is not allowed here on the English Wikipedia, the WP:NOLLM guideline makes it impermissible. ‑‑gurkubondinn 17:32, 12 July 2026 (UTC)

Ok kindly yourself type and add the contents of Meivazhi Incorruptibility so it be know to all, so people will find out the by debunking incorruptibility or study, so truth will come out. Kindly help me if facts I stated in meivazhi content is very true, it's happening & it's truly baffling Wisdomsalai (talk) 17:39, 12 July 2026 (UTC)

for your kind information, since 1933, all Meivazhi followers attain divine incorruptibility as I stated with those observable physical signs, so I want this recorded & people should know about it.. I'm not a native English speaker, kindly help me post the contents in wikipedia about Meivazhi incorruptibility Wisdomsalai (talk) 17:43, 12 July 2026 (UTC)

If you can't make the edits yourself then you should make an WP:EDITREQUEST instead. Unfortunately, I have no interest in this subject so I can't help you with making these edits. The problem with the edits that you made is that they were generated by an AI chatbot, which is not allowed. ‑‑gurkubondinn 17:48, 12 July 2026 (UTC)

You can also ask for help in the WP:TEAHOUSE, there is a lot of friendly people there that might be able to help you out. You could also try asking at WT:WikiProject Catholicism, but I don't know how active it is. ‑‑gurkubondinn 17:49, 12 July 2026 (UTC)

pls kindly help me, I'm not good with technology.. I wanted just to post this article & let world know, so kindly I'm asking you to post it yourself or edit my content place it.. I thought wikipedia was the right place to access right information about Meivazhi. Pls i don't know to how to post it so pls.. kindly help me post the contents of Meivazhi Incorruptibility Wisdomsalai (talk) 18:01, 12 July 2026 (UTC)

my internet network is rural part of South India, kindly help me restore the contents in the page, you edit and post it according to Wikipedia guidelines, all the things I stated is true to best of my knowledge & i have found about them after long struggles. Do kindly help me post the page in incorruptibility page Wisdomsalai (talk) 18:08, 12 July 2026 (UTC)

No, I have already told you that AI-generated text is not allowed on the English Wikipedia. Everything also needs to be cited to wp:reliable sources. If you want help, then the wp:Teahouse is the right place to ask. ‑‑gurkubondinn 18:13, 12 July 2026 (UTC)

Hi! You mentioned in a discussion a no-Wiki that you do a lot of cleanup after LMM users. If this also include cross wiki spamming of machine translated articles to languages the user has no knowledge og, then m:Spesial:GlobalContributions/MaryTheo2012 is a candidate to follow.

Blocked at no-Wiki, have had content deleted or moved to user pages on several language versions (see also User talk:MaryTheo2012), repeated removal of the template "maskinöversatt" at sv-Wiki (I have now escalated to speedy delete). 11 uploads at Commons, all of them copyvio (of which 7 are already deleted). 1000mm (talk) 23:44, 13 July 2026 (UTC)

Well... yeah, that's ridiculous. I saw the edit warring over the "maskinöversatt" tag on svwiki, and this person doesn't seem to speak a word of Swedish like you pointed out over there. This person doesn't speak a word of Icelandic either, because is:MCND is clearly not made by someone that understands the language, so I'll see about getting that deleted. Thanks for blocking the user on nowiki. ‑‑gurkubondinn 10:04, 14 July 2026 (UTC)

Somewhat related, but svwiki's policy on machine translation is in my opinion much more reasonable than the enwiki WP:LLMT guideline;

Maskinöversättning bör inte användas annat än som hjälpmedel. För att skriva artiklar på svenskspråkiga Wikipedia måste man behärska svenska språket.
— sv:Wikipedia:Raderingsförklaringar § Maskinöversättning

‑‑gurkubondinn 10:15, 14 July 2026 (UTC)

I've opened WP:AINB § User:MaryTheo2012. ‑‑gurkubondinn 11:20, 14 July 2026 (UTC)

what's your reasoning behind engaging heydraist like this instead of sending them to ANI? ~2026-38678-64 (talk) 22:43, 15 July 2026 (UTC)

Someone else can take them to ANI, this user will probably find themselves there sooner or later. I don't like filing at ANI if I don't have to, and there's no "urgent and persistent" disruption going on anyway. Why not hand out some ROPE and take the answers given to their logical conclusion? ‑‑gurkubondinn 22:58, 15 July 2026 (UTC)

@~2026-38678-64, and because of stuff like this: Diff/1364331301. Guess what the reply included? Because of course it did: Diff/1364332221. ‑‑gurkubondinn 23:58, 15 July 2026 (UTC)

That was hilarious. Thanks, by the way. PARAKANYAA (talk) 00:16, 16 July 2026 (UTC)

Thanks for bringing this to AINB. See, problems can be solved without the dramaboard :) ‑‑gurkubondinn 00:20, 16 July 2026 (UTC)

Yeah, it wasn't urgent or persistent, but I thought it best to nip it in the bud before it becomes a cleanup problem on more than 1 article. PARAKANYAA (talk) 00:34, 16 July 2026 (UTC)

i continuously forget that admins regularly patrol AINB ~2026-38678-64 (talk) 00:40, 16 July 2026 (UTC)

Hi Gurkubondinn, I just wanted to let you know that I have added the new page reviewer user right to your account. This means you now have access to the page curation tools and can start patrolling pages from the new pages feed. If you asked for this at requests for permissions, please check back there to see if your access is time-limited or if there are other comments.

This is a good time to re-acquaint yourself with the guidance at Wikipedia:New pages patrol. Before you get started, please take the time to:

• Add Wikipedia talk:New pages patrol/Reviewers to your watchlist to follow NPP-related discussions
• If you use Twinkle, configure it to log your CSDs and PRODs
• If you can read any languages other than English, add yourself to the list of reviewers with language proficiencies

You can find a list of other useful links and tools for patrollers at Wikipedia:New pages patrol/Resources. If you are ever unsure what to do, ask your fellow patrollers or just leave the page for someone else to review – you're not alone! _signed, Rosguill ^talk 23:44, 15 July 2026 (UTC)

Thanks, @Rosguill! ‑‑gurkubondinn 23:49, 15 July 2026 (UTC)

Thank you for your feedback. Iam junior here but I will learn soon i hoop.

The draft has been rewritten manually to ensure it reflects a neutral, source-based encyclopedic style. If there are specific sentences or sections that appear problematic, I would appreciate it if you could point them out so I can revise them accordingly. Sajjadmrt (talk) 20:24, 17 July 2026 (UTC)

Hi, there's nothing wrong with being new here. We are all new to this at some point. Did you use a tool like ChatGPT or an "AI" to help you with your draft? ‑‑gurkubondinn 20:36, 17 July 2026 (UTC)

Thanks for your time. Yes Iam so sorry. For lead I used. I didn’t the rules. Never happened. I try best for that draft. Sajjadmrt (talk) 21:17, 17 July 2026 (UTC)

That's fine, mistakes happen. You didn't know, and that's hardly your fault. There is a lot of rules, guidelines, and policies around here and it takes some time to absorb all of it. There are some guides that I think that you may find helpful;

• WP:YFA: writing your first article
• WP:REFBEGIN: referencing for beginners
• WP:NOLLM: we are not allowed to use AI or LLMs (things like ChatGPT) on English Wikipedia

These are a bit more lengthy and in-depth guides to help you getting oriented around here;

• WP:INTROLIST: a list of tutorials
• WP:HOW: contributing to Wikipedia, and how it all works
• H:INTRO: an introduction to Wikipedia

I hope this helps, ‑‑gurkubondinn 21:23, 17 July 2026 (UTC)

I am trying to carefully avoid canvassing here, but I wish to ask if you could look at the dispute between Kaspar-trout and I at ANI and simply give an opinion as to which one of us is misinterpreting or misrepresenting the diff. And if it is me, please feel free to dump an entire cannery line of trout on me, because I will have deserved it. M kuhner (talk) 21:49, 17 July 2026 (UTC)

You're the third person to ask me about this user, that has to be some kind of a DUCK variation or something. Personally I think that the premise of the ANI filing is entirely correct, but this might be a WP:DEADHORSE regardless. Sometimes it's just best to peace out and see if someone else picks up the metaphorical stick instead. And some editors will just inevitably find themselves back at ANI sooner or later, so may as well just let that happen. Getting drawn into disputes is not worth it (and WP:CGTW imparts a lot of good points).

You're still not wrong though. ‑‑gurkubondinn 22:24, 17 July 2026 (UTC)

I mainly wanted to check I wasn't making a ghastly interpretation error. Thanks for the second opinion, and I'll take your advice and just quit looking at that one.

M kuhner (talk) 00:24, 18 July 2026 (UTC)

hi, gurkubondinn! I'm a chronic browser of the noticeboards and a fairly sporadic editor (been using IPs for a few years, mainly typos and stuff.)

I've been lucky enough to have a lot of free time lately and over the past half a year or so I've noticed you and Gnomingstuff REALLY putting in the work on LLM/AI related crap. I work very closely with LLMs due to the industry I am in as a project manager. Is there any way I can help you? What could I do that would be the most effective in order to at least TRY to ease the ever growing tide of slop and lies? Pardon any typos please, I'm mostly on mobile. Hooples (talk) 13:39, 19 July 2026 (UTC)

Just hang around the WP:AIN noticeboard and WT:AIC talk pages, would probably be a good start. It's mostly better to use a proper computer and keyboard though, because this involves a lot of looking at a million different diffs and copying an ungodly amount of text around. There's some guides and stuff, I think of most of them can be found on the WP:AIC project page. I think everyone finds their own ways of identifying this stuff. Personally, I look a lot at some of the edit filters, and I have a bunch of searches that I look at. I mostly think of it in terms of users, not in terms of articles. So when I find something somewhere, I take a look at the user and quite often that uncovers a lot more. And looking at timestamps, it's very often obvious just from the sheer speed and magnitude.

There's a whole bunch of open cases at AIN. You can always pick any of them (if you ask me which ones, I'd just point you to some of the larger cases that I have filed and haven't gotten finished), if you want to get into actual cleanup rather than identification. But for that I definitely recommend using a proper computer.

A lot of this also comes down to (very patient) user communication. You have to be good at not biting the (very delicious) newcomers that simply don't know better. You'd quickly learn to tell the clueless apart from the maliciously intended (this is actually not that different from the pre-ChatGPT days, the difference is just in volume). It's a lot communication that goes into this.

There's also the policy discussions. There's an open RfC at WT:NOLLM right now, and one more somewhere else. I used to keep the list at the AIC page updated, but I've kind of forgotten about that lately, and I'm not sure how up to date it is.

You can also watchlist my talk page, and feel free to jump in when someone shows up here. I have a few talk page stalkers by now, and I'm always glad when they show up (to stalkers reading this; you are welcome, and you are appreciated). ‑‑gurkubondinn 13:47, 19 July 2026 (UTC)

Thank you for the wonderfully helpful words, and I'll take on that advice. Shame I have to get on my laptop, eh? For those who are reading and may get a kick out of this – the reason I have time to get back into wiki at all is that a project I am working, a new data center (gasp) has been deadlocked by a county declining the permits for construction to begin. I'm happy about this both because fuck AI and I have a lot of free time for a month or so, but my wallet sadly isn't. I did not know about the RFC! I will do further reading on this lazy Sunday. Thanks again :) Hooples (talk) 13:53, 19 July 2026 (UTC)

Just FYI, you reverted my edit removing the redlinked categories from Mahmoud Kafil Uddin and restored the edit by the editor who was blocked for using LLM. I'm pretty sure that's not what you intended if the LLM was the problem. I've already restored the last clean edit, so no action needed, but just wanted to let you know. Bearcat (talk) 15:50, 19 July 2026 (UTC)

Thanks! I hadn't gotten to the second check of that article yet. I normally go back and double check that I restored the correct revision (I sometimes click wrong..). ‑‑gurkubondinn 15:54, 19 July 2026 (UTC)

Administrator Elections | RFC phase

The RFC phase of the May 2026 administrator elections has started. There are 20 RFCs for consideration. You can participate in the RFC phase at Wikipedia:Administrator elections/May 2026/RFC phase.

Any questions or issues can be asked on the election talk page. Thank you for your participation. Happy electing.

You're receiving this message because you signed up for the mailing list. To opt-out of future mailings, please remove yourself from the list.

MediaWiki message delivery (talk) 16:02, 20 July 2026 (UTC)

Hi gurkubondinn. Thanks for such an interesting home page, and my apologies for being impertinent enough to make two minor typo edits on it. No disrespect intended. Sometimes I just can't help myself. All the best. Dropbear2 (talk) 22:32, 20 July 2026 (UTC)

No worries at all, thanks for fixing the typos! Glad you found it interesting, because it's mostly a collection of stuff that I find interesting. :) ‑‑gurkubondinn 08:31, 21 July 2026 (UTC)

Hi Gurkubondinn. You have redirected the Jamalpur High School to Jamalpur Zilla School#History. But both schools are different and the high school has nothing to do with the history of the zilla school. I would suggest deleting the high school's page rather than making an unreasonable redirect. ~ Sàádî ^✉️ 10:25, 21 July 2026 (UTC)

The school was placed under government pay roll in 1882 and renamed Jamalpur Dono English High School.

Is that not referring to the school that the AI-generated article was about, or did I misread/misunderstand this? The AI-generated article mentioned the Zilla school as well, but I'm not sure how reliable that is (it seemed a bit out of place as well). ‑‑gurkubondinn 10:30, 21 July 2026 (UTC)

Maybe retargeting the redirect to Board of Intermediate and Secondary Education, Mymensingh or Jamalpur District would work? Unfortunately neither of them mention the school though. ‑‑gurkubondinn 10:32, 21 July 2026 (UTC)

I've retargeted the redirect in Diff/1365275617 to the article for the Board of intermediate and secondary education. ‑‑gurkubondinn 10:45, 21 July 2026 (UTC)

Also, the Board of Intermediate and Secondary Education is the controller of all schools and colleges in this administrative division. I think just one school does not represent it well. ~ Sàádî ^✉️ 11:16, 21 July 2026 (UTC)

I agree, it's not a representation of the board. But since the school is managed by the board, I think that the name of the school is redirect to the board (as an {{r with possibilities}} redirect) until someone writes an article about the school. ‑‑gurkubondinn 13:43, 21 July 2026 (UTC)

Another possible redirect target could be list of schools in Bangladesh, but it doesn't have a section for the Mymensingh. There is a section § Dhaka Division, which does list some schools in Mymensingh, but as far as I can tell, Jamalpur District is not a part of the Dhaka Division since 2015 when the Mymensingh Division was split out into its own division? ‑‑gurkubondinn 13:53, 21 July 2026 (UTC)

I found List of universities and schools in Mymensingh it could be fit here. And I am creating the missing section of Mymensingh Division on list of schools. Sàádî ^✉️ 14:22, 21 July 2026 (UTC)

Oh, nice. I tried looking for that first, but I didn't find it so I thought that it didn't exist. My mistake, but that is definitely a better redirect target. ‑‑gurkubondinn 14:23, 21 July 2026 (UTC)

Jamalpur English High School was the earlier name of the Zilla School and yeah, it's not the same as the Ai one. Also, I looked up the citations given in the Ai-generated one (about the Zilla School), but I found nothing. Sàádî ^✉️ 11:11, 21 July 2026 (UTC)

Hello @Gurkubondinn Hello, Hope this finds you well. I wanted to ask why the changes for the Genta Ismajli article were removed and reverted when I was trying to help and contribute on the article? Please let me know here on what the issue appears to be. Thank you in advance. ~ Pulseoraus (talk) 21:46, 22 July 2026 (UTC)

I wrote it in the edit summary. There was an WP:IMDB reference, and I think it was LLM-generated. ‑‑gurkubondinn 21:48, 22 July 2026 (UTC)

It is not. How come you're flagging all of the articles I have spent time on contributing as AI or LLM? I understand you're very against it but that is not an accurate course of action. Even the desnudo album I worked with independent editors who fixed it. I see you hate AI and so do I. please stop making false accusations on every article I've worked on.

~ Pulseoraus (talk) 21:51, 22 July 2026 (UTC)

Because I see very strong signs that they have been generated by ChatGPT in all of them. And we don't need two discussions. Please keep it to the one on your talk page. ‑‑gurkubondinn 21:54, 22 July 2026 (UTC)

Hi, I noticed you had restored the LLMPROD tag to 2024 European Archery Championships – Men's individual compound (since deleted) after I had previously checked it over. Can you please explain what you mean by "WP:LLMPRVOBJ has not been followed"? and "The Special:Diff/1358414589/1359551650 changes are not sufficient"? I had stated in the edit history that I had "checked, seems fine", and that I can "take responsibility for it". LLMPROVBJ only requires the article be "reworked as needed", which I did. ARandomName123 (talk)^{Ping me!} 11:26, 23 July 2026 (UTC)

Basically, I just didn't considered it to have been reworked "as needed". The LLMPRVOBJ requirements are actually intentionally written to be pretty "high effort", and just saying "I don't see a problem" isn't enough. I think most of the text still remained on the article and you had just changed/removed/added some references? It was still largely the same text that the (now blocked) LLM user had inserted, stubifying it might have been better (but I can't see the article myself either, so I'm just going off memory). I was just going through my notes of the cleanup and discovered that page, I have no strong feelings about this page in particular or anything. But if you want to request a REFUND, then I wouldn't oppose it in any way. ‑‑gurkubondinn 11:31, 23 July 2026 (UTC)

I undeleted the page. I hadn't changed any text because there's almost no text to be reworked. There's only like four sentences in the lead, the rest is just tables. Even if there was additional text, the fact that I have taken responsibility for it means that LLMPROVBJ applies. It shouldn't matter that "it was still largely the same text", as long as the text has been checked. ARandomName123 (talk)^{Ping me!} 11:39, 23 July 2026 (UTC)

Well, but it does. Just "taking responsibility" for it doesn't really help, because that just means that you are "taking responsibility" for the NOLLM vios. There are no allowances or carveouts for "LLM-generated text that has been checked", so merely checking the text doesn't "fix" the problem. It like how you can't "take responsibility" for BLP vios, or for spam, etc etc. But yes, these sports articles that are 90% tables are not the easiest to deal with, you'd either have to remove the table or carefully check that everything in the table is correct (which not difficult, just super tedious).

They're actually very tedious, because for a minute I thought that there was a source integrity issue. Theyre hard to read, it's easy to misread it as Mike Schloesser having won the tournament and not placed third (like the dsb.de source says). But the leade spells it out correctly, just meant that as an example of how these tables are fiddly and tedious to parse (so I'm not entirely convinced of their usefulness but thats a different question). I would still prefer the LLM-generated text to be rewritten, but you are right that there isn't a whole lot to rewrite and it is mostly just stating things in ways that are difficult to rephrase (so not the biggest issue).

This user has been returning with TAs, and has edited some of the remaining archery articles. It hasn't happened for a few weeks now, but it's something to keep an eye out for on this article. ‑‑gurkubondinn 11:51, 23 July 2026 (UTC)

A listing table like in § Qualification round would probably be more useful than the confusing rounds tables in the § Elimination round sections. But that's coming from someone who knows approximately nothing about archery, so I may be completely wrong. ‑‑gurkubondinn 11:54, 23 July 2026 (UTC)

Also what does 'Bye' in the tables mean? DDG wasn't very helpful, but I guess it is an archery (or sports?) term? ‑‑gurkubondinn 11:56, 23 July 2026 (UTC)

LLMPRVOBJ states that the editor restoring it "may assume responsibility for the content". I was under the impression that the way you objected to an LLMPROD was to check it, make fixes as necessary, then state that you have done so by taking responsibility for it. You can also state that you take responsibility for an edit that adds a BLP vio, it's just that you'll be warned and probably blocked for it.

I also know next to nothing about archery (I was trying to see how hard it would be to save an article from LLMPROD), so I just compared it to the other archery articles (ie. 2022 European Archery Championships – Men's individual compound) and saw they had the same structure.

Per Merriam 'Bye' just seems to mean that a competitor is able to advance without competing in that round. ARandomName123 (talk)^{Ping me!} 12:07, 23 July 2026 (UTC)

• Draft:John Henry Schwarz (edit | talk | history | links | watch | logs)

Interesting case. The prose itself seems human-written to me (and Pangram and GPTZero agrees), but clearly LLM was involved in some way given the dead refs and markdown. Could be a case of LLM formatting just introducing hallucination to otherwise human text. Ca ^{talk to me!} 00:00, 24 July 2026 (UTC)

Yeah, and I find the ==References== {{reflist}} part odd as well (and I think that there is some family-level COI as well because of the username). An odd number of references also point to redirects, which wouldn't happen if you copied and pasted the URL of a website that you were looking at. I hadn't checked before, but wikipedia.gptzero.me gives the scores ${\displaystyle 0.963}$ (human), ${\displaystyle 0.028}$ (ai) and ${\displaystyle 0.009}$ (mixed). Could very well be a false negative, but only the submitting user themselves can tell for sure. ‑‑gurkubondinn 09:09, 24 July 2026 (UTC)

Looks like it's been deleted by User:BusterD. Since the text itself appears to be plausibly human-written, what do you think of restoring the draft and asking the author about their LLM use? I'll keep the draft watchlisted. Ca ^{talk to me!} 11:41, 24 July 2026 (UTC)

What about asking them first? I'm not super keen to rely on a algorithmic classification to determine if something was algorithmically generated or not, talking to the user seems like a better approach to me. If they're interested in working on the draft, then it could be restored for them. ‑‑gurkubondinn 12:37, 24 July 2026 (UTC)

I'd be glad to restore if you two think we can win ourselves a new working wikipedian. Please feel free to ping me, but if I'm not immediately available, I'd be fine with User:Ca (or other admins) restoring, so long as the situation is supervised. I often delete User:Gurkubondinn's tagged pages, and find myself agreeing with their analysis. BusterD (talk) 22:04, 24 July 2026 (UTC)

Just making you aware of Draft:ETIR International System and Draft:International TIR Data Bank as well. Cheers and thanks for your work. Sulfurboy (talk) 19:51, 24 July 2026 (UTC)

Yeah, checked those. Don't meet G15 criteria, so they just need to be tagged (I also leave the reason in the edit summary where it can't just be removed) and rejected so they don't end up in article space. ‑‑gurkubondinn 15:12, 25 July 2026 (UTC)

Just reverted the same user on eTIR, there is a markdown thematic break in Diff/1365794437. ‑‑gurkubondinn 16:48, 25 July 2026 (UTC)