Jump to content

Wikipedia:AI noticeboard/Archive 13

Add links
From Wikipedia, the free encyclopedia
< Wikipedia:AI noticeboard
Latest comment: 14 days ago by M kuhner in topic User:Guywelch2000
Archive 10Archive 11Archive 12Archive 13Archive 14Archive 15Archive 16

User:Billyjeanheheehe

Latest comment: 1 month ago1 comment1 person in discussion

The Billyjeanheheehe user is an alternative account of Kyleoahsom, who is a been a previous guest here (courtesy of Fermiboson: Wikipedia:AI noticeboard/Archive 9#User:Kyleoahso.

Found the Billyjeanheheehe user because they created the Magsaysay Drive (Olongapo) article with a markdown ``` code block left over from their chatbot. ‑‑gurkubondinn 15:47, 10 July 2026 (UTC)

User:Bondwagon

Latest comment: 1 month ago8 comments7 people in discussion

Yet another serial AI user. The reason I am making a noticeboard thread here is that they have a charming habit of removing the tags on their obvious (and eventually disclosed) AI articles with edit summaries like "typo". Gnomingstuff (talk) 03:35, 3 July 2026 (UTC)

Bondwagon hasn't edited at all since April and the actions being commented on here mostly occured in 2025. This tag was removed in August 2025 with the edit summary "removed likely to be AI tag". This tag was removed in October 2025 with the edit summary "typo". Gnomingstuff restored it.
I still see signs. SenshiSun (talk) 05:21, 3 July 2026 (UTC)
April is only 3 months ago. I am opening a noticeboard thread so there is documentation of this and I don't have to wait until the next time I stumble across the same article because the same shit has not gotten fixed and get greeted with the tag being gone with the mocking-ass edit summary "typo". Gnomingstuff (talk) 05:54, 3 July 2026 (UTC)
I like when in April 2025 they said My additions aren't AI generated. They are a result of careful research , and then the next month they started adding utm_source=chatgpt.com , careful research indeed. Later on they stated that AI only did spelling and grammar check .
These "spelling and grammar checks" resulted in classic generated text like ... an important tourist destination in Rivers State, known for its cultural heritage and historical significance , The vibrant community of Anyama persists in its trading activities, specializing in a diverse array of ... , The ceremony was marked by vibrant displays of color, with guests dressed in matching ... or Kaiama is distinguished by its swampy terrain and lush vegetation. Unlike other ... .
It's clear that this editor has a history of AI use in a generative capacity, and so their edits may be removed via 1B of WP:LLMPRV. fifteen thousand two hundred twenty four (talk) 06:25, 3 July 2026 (UTC)
Bondwagon has requested their deleted pages be restored to their userspace at WP:REFUND, ironically with a request that itself appears to be AI-generated. * Pppery * in solidarity 19:41, 14 July 2026 (UTC)
Went and tagged all of the restored pages because I'm afraid they'll just find their way back into article space, and we'll just have to start all over. Funny how they admitted to using an LLM in the refund requests. ‑‑gurkubondinn 20:26, 14 July 2026 (UTC)
the requests are at Wikipedia:Requests for undeletion/Archive 424, they haven't edited since but worth keeping an eye on Kowal2701 (talk, contribs) 15:39, 24 July 2026 (UTC)

Made a tracker from January 2025, which seems to be the inflection point. Not positive, though. M kuhner (talk) 22:41, 30 July 2026 (UTC)

User:Ertuyotuyo

Latest comment: 1 month ago13 comments7 people in discussion

Warned for AI usage by other people; says they're only using it for formatting, but is creating articles at an incredibly fast clip (and removing tags). Gnomingstuff (talk) 23:31, 5 July 2026 (UTC)

Reference 5 in Almanaque Panayanhon only contains the name in a citation, “Mga Magsusulat sa Panay Cutub sang Tiempo sang Cachila Tubtub Caron.” Almanaque Panayanhon,
1914. Mandurriao, Iloilo City: Imprenta y Libreria Panayana de Casimirio F. Perfecto,
1913. 66. Impreso. SenshiSun (talk) 00:06, 6 July 2026 (UTC)

Good day! Our team already has a lot of articles that we post on our Facebook page, Historia Antikenya. We decided to publish these articles on Wikipedia and have been using a generator tool to format them, as we have no idea how to code. We sincerely apologize if we have caused any issues. We had a lapse in checking whether the bot added any false information or citations. However, to back up our integrity, please know that we research our articles and write the information ourselves. Ertuyotuyo (talk) 03:20, 6 July 2026 (UTC)

Moving forward, we will no longer use the generator tool. I am actively learning how to format using Wikicode myself. Please note that following yesterday's warnings, I have already gone back through our previous uploads to remove any false information and correct the formatting issues. Ertuyotuyo (talk) 03:27, 6 July 2026 (UTC)

Kowal2701 (talk, contribs) 09:41, 6 July 2026 (UTC)
This is far more serious than AI cleanup. This is copyvio cleanup, which is difficult in this case due to language translation. The Wikimedia Foundation has not been given permission by the copyright owner to republish content that is already published on Facebook. ~Anachronist (who / me) (talk) 14:58, 6 July 2026 (UTC)
It also appears that the account is being shared (note the repeated 'we' above), contrary to policy. And I'd question whether, regardless of copyright issues, the material from this 'team' meets WP:RS criteria. AndyTheGrump (talk) 16:53, 6 July 2026 (UTC)
Yes, they are multiple people and deserve multiple blocks, indef in fact. Yesterday, all my dreams... (talk) 11:18, 7 July 2026 (UTC)
See Ertuyotuyo's talk page explanation. I am satisfied the account is a single person. The editor has also agreed to refrain from further contributions until the WP:CONSENT process is finished. ~Anachronist (who / me) (talk) 16:11, 7 July 2026 (UTC)
  • Excuse me, but why is there so much tolerance towards these disruptions? Why is there no admin action? Yesterday, all my dreams... (talk) 11:21, 7 July 2026 (UTC)
    Because none is needed after communicating with the editor, which some participants should try doing. This is a collaborative project that requires communication. Maybe you should try communicating with newbies who are editing in good faith, not knowing what they're doing wrong. In this case, it's a new user who is acting in good faith, not using AI except for translation of original articles written by that editor and associates, and who was unaware of the AI and copyright issues and wants to do the right thing. ~Anachronist (who / me) (talk) 16:17, 7 July 2026 (UTC)
    The user (or one of them) started in 2024 ! Not new. Again the keyword is: timesink. Collaborative has to be a 2 way street. For the AI addicts it is not. Yesterday, all my dreams... (talk) 17:01, 7 July 2026 (UTC)
    The start date doesn't matter. The low edit count suggests inexperience. That is the sense in which I used "new". This user responds to comments, and is therefore collaborative. A block is intended to prevent further disruption, and if an editor has stated that they need guidance in doing the right thing, then no reason has been offered by anyone here to issue a block. ~Anachronist (who / me) (talk) 00:53, 8 July 2026 (UTC)
    Anyway, I am going to stop looking at this board. The AI material is getting into Wikipedia like Marine plastic pollution and the few editors here will eventually be overwhelmed and pollution will sink into the encyclopedia. Sad, but inevitable. C'est la vie. Yesterday, all my dreams... (talk) 20:15, 7 July 2026 (UTC)

From July 6 edit, subsequent to the warnings: Gella belonged to the prominent Gella family of Antique. His brothers included Vicente Gella, who served as governor of Antique; Ariston Gella, the province's first pharmacist and a delegate to the Malolos Congress; and Bartolome Gella. He was married to Baltazara Yturralde, a member of another prominent provincial family. Sourced to which does not mention Ariston or Bartolome, does not mention Malolos or delegate, and mentions a Vicente but with a different family name. This looks hallucinated. We need some kind of cleanup action, as the editor does not seem to be cleaning up despite continuing to edit, most recent edit July 18. (Very recent edits are gnoming and cannot be checked for LLM, but they are not cleanup.) M kuhner (talk) 19:48, 26 July 2026 (UTC)

Jacob6493

Latest comment: 1 month ago4 comments3 people in discussion

Behavior I cannot explain besides with LLM usage.

~ A412 talk! 04:44, 8 July 2026 (UTC)

I draftified the page creations and left a warning on his talk page. The mainspace edits do need to be cleaned up though. ~Anachronist (who / me) (talk) 05:39, 8 July 2026 (UTC)
@Anachronist, just FYI, imo it's best to userfy articles (or tag drafted ones w {{AI-generated}}), I keep seeing draftified articles get moved back into mainspace much later by someone else whose unaware of the context (eg. ) Kowal2701 (talk, contribs) 22:15, 16 July 2026 (UTC)
If you see that happen, please make the person aware of the context. In the example you linked, the person who moved it to mainspace is an administrator, and should have known why it was draftified in the first place. ~Anachronist (who / me) (talk) 22:29, 16 July 2026 (UTC)

User:SunriseInBrooklyn

Latest comment: 1 month ago4 comments2 people in discussion

SunriseInBrooklyn (talk · contribs), XTools

Was blocked June 2024 for CIR and AITALK at Wikipedia:Administrators' noticeboard/IncidentArchive1159#GA-banned user violating topic ban. Was unblocked June 2025 where they said Regarding AI: I now understand using such as ChatGPT or any similar tools without through human oversight violates community standards. I regret doing so and will not AI for editing again. I. Received a warning from @Gurkubondinn in 24 June 2026 (they haven't edited since). 9,700 edits (no relevant article creations). Kowal2701 (talk, contribs) 16:54, 9 July 2026 (UTC)

To any patrolling admins: is this worth a mainspace block? They've had a lot of rope, and I'm concerned that they're just going to return at some point and carry on. (btw courtesy ping to 331dot as the unblocking admin, no need to respond if too busy) Kowal2701 (talk, contribs) 14:41, 11 July 2026 (UTC)
This also reads as probably AI-generated to me:

... has a long and varied history within ..., the street evolved over time ..., reflecting its intended role as a major thoroughfare. ..., hinting at its ambitious design origins.
— Diff/1304419918

There's also an utm_source=chatgpt.com in there, but I don't want to read too much into that alone. ‑‑gurkubondinn 15:53, 11 July 2026 (UTC)
Back in August 2025, they triggered the 702 (log) edit filter ('Warning against clipboard hijacking') twice in row: AbuseLog/41839229 and AbuseLog/41839237, saved as Diff/1304545542. So they must have copied that text from somewhere. ‑‑gurkubondinn 16:00, 11 July 2026 (UTC)

LLM use by IyyasTihlah socks

Latest comment: 1 month ago3 comments3 people in discussion

An ANI case WP:ANI § AI editing by User:TihlahAbdellah started with LLM concerns and ended with the discovery that they were a sock. All sock edits (three accounts worth) have been reverted, but the master IyyasTihlah589 has a small number of edits and two article creations that should be checked for LLM. M kuhner (talk) 05:10, 15 July 2026 (UTC)

I tagged one article as an obvious G15, but the other one Dar Al-Thaqafa is too short to tell; it's a reasonable stub. ~Anachronist (who / me) (talk) 05:24, 15 July 2026 (UTC)
It's been significantly rewritten, the original version was pretty unambiguous Gnomingstuff (talk) 07:56, 15 July 2026 (UTC)

User:Deedubaroo

Latest comment: 1 month ago39 comments9 people in discussion

Could I get a second opinion on my tagging of the Regulation to Prevent and Combat Child Sexual Abuse page? First a TA removed the tag in Diff/1364598177, and after i restored the tag in Diff/1364609031 it was again removed without consensus in Diff/1364613370 with an edit summary calling it an overbroad AI tag.

The user has confirmed using LLMs in § Oh darn, you're right of course on my talk page, but evidently does not like the tag. There's too much of the usual arguments about NOLLM not really applying and what it all means etc., so I'm asking for a second opinions here. ‑‑gurkubondinn 16:09, 17 July 2026 (UTC)

Just my 2 cents. I have repeatedly acknowledged the policy after being made aware of it, spent over an hour reviewing the page and the policies, and subsequently reaffirmed a commitment to the policy. The tag was applied blindly and mechanically in a manner that (as documented) clearly violates Wikipedia policies.
Without regard for adopting a more considered approach to tagging, or justifying their edits by reference to one of a list of principles associated with the rule (i.e. acting without effort in a manner that destroys effort), both of which were suggested to the tagger, or commonsense investigation of the quality of the page before or after, they have instead responded with a further counter-productive escalation ala. WP:BUREAU. It is my view this is WP:DE in the context of the particular page, although I continue to reaffirm and support both the LLM policy and the overarching principle on which the tagger is operating. Deedubaroo (talk) 16:21, 17 July 2026 (UTC)
The tag was not "applied blindly". Special:Diff/1364569730 contains signs of probable AI use, such as markdown formatting (which is not present in prior edits so it's not due to unfamiliarity with wiki markup). As it turns out, those signs were correct, which moves it from "probable AI use" to confirmed AI use. Thus, the tag is accurate, and removing it misleads both editors and readers. If you don't like the tag, don't use AI, and then ask yourself why you are using a tool that you are ashamed of people knowing about.
As for the spirit of the underlying policy, both the letter and spirit of the guideline are that the Wikipedia community does not want people to use AI here. The principle is to not use AI. Gnomingstuff (talk) 17:04, 17 July 2026 (UTC)
Hello, the template applied states: "As is the case for most maintenance tags, this should only be removed once the problems are undetectable or demonstrably non-existent." I have:
  • previously included reference to the list of defects associated with the tag.
  • already confirmed the page has been reviewed
Can someone please point to detectable signs of AI usage in the current version of the page, i.e. "factual errors", "formatting issues", "fictitious but plausible-seeming references", or prose that is "stilted", "promotional", or "occasionally ludicrous". Deedubaroo (talk) 17:13, 17 July 2026 (UTC)
Ignoring the fact that I already pointed out formatting issues above, the most detectable sign of AI usage at this point is that you literally said that you used AI. No more detection is needed. Gnomingstuff (talk) 17:20, 17 July 2026 (UTC)
You are continuing to engage with a blind and mechanical view of the relevant policy, which clearly states:

Text generated by large language models (LLMs) often violates several of Wikipedia's core content policies. For this reason, the use of LLMs to generate or rewrite article content is prohibited

It continues:

When evaluating possible LLM use, it is best to consider the full pattern of the editor's recent edits and whether the edits comply with core content policies.

The full pattern of the user's edits is clearly a manual labour poured in across more than a week. This requires a degree of nuance that is absent from the rule itself - the principle that the LLM has not been used here to create a problem of the sort the rule is intended to prevent.
You have offered no substantive reason despite multiple requests to suspect the page in question has quality problems in its current version, have not indicated how it violates content policies, and have not acknowledged commonsense requests to review its quality before and after. Simply acknowledging the use of AI is productive and honest behaviour done for the purposes of returning to editing as soon as possible. Policing without cause as done here however, is not.
Your mandate to police Wikipedia is based on implementing its content policies, not to police for policing's sake. This is pointless counterproductivity. A common sense view of the no-LLM policy is that it is designed to catch flagrant cases where people are generating entire pages filled with hallucinations and interpolated facts without any effort or human review. That ought to be what use of this tag helps correct, not momentary syntax errors in otherwise reviews content (which is plainly obvious by the delay between the flagged diff and the subsequent manual correction of errors). Deedubaroo (talk) 17:41, 17 July 2026 (UTC)
The "substantive reason" is that AI-generated text is not allowed on Wikipedia. That is the violation of Wikipedia's guidelines. You can complain about the guidelines until you are blue in the face -- or use AI until you are blue in the face, given that you have suddenly started to refer to yourself in the third person (The full pattern of the user's edits) -- but that doesn't make you an exception to them. Gnomingstuff (talk) 17:52, 17 July 2026 (UTC)

For this reason, the use of LLMs to generate or rewrite article content is prohibited

This behaviour ceased the moment it was originally flagged, i.e. the violation of the prohibition has stopped, not continued, ended, whatever you need.
Ad hominem isn't useful. Of course I'm not using AI to edit these. If I had been, the above comment wouldn't be so full of typos. I'd expect a well trained officer of the law to recognise such things; that you do not explains this whole issue. Also as a reminder, Wikipedia's rules are principles, not civil code or exacting law, in spite of which WP:Wikilawyering is clearly a viable profession around these parts. Deedubaroo (talk) 17:58, 17 July 2026 (UTC)
I'd expect a well trained officer of the law to recognise such things
Good thing I'm not a "well trained officer of the law," then. The only person wikilawyering here is you.
This behaviour ceased the moment it was originally flagged, i.e. the violation of the prohibition has stopped, not continued, ended, whatever you need.
Which only means that you're no longer violating the guidelines. It doesn't mean that the text generated by AI is somehow grandfathered in. Gnomingstuff (talk) 18:06, 17 July 2026 (UTC)
For the umpteenth time, and I notice someone else has requested this too, please indicate how the current content of the page violates Wikipedia content guidelines. You are most welcome to compare it to the old link page with respect to WP:NPOV in particular Deedubaroo (talk) 18:11, 17 July 2026 (UTC)
That's easy. It was generated by an LLM, and that means that it fails NOLLM. This been pointed out umpteen times already. ‑‑gurkubondinn 18:15, 17 July 2026 (UTC)
And sentences like The proposal allowed a judicial or administrative authority to issue a detection order where mitigations had not ... from Diff/1364569730 still present in the article, meaning that there incontrovertibly is LLM-generated text still in the article and that the problem has not been addressed. ‑‑gurkubondinn 17:26, 17 July 2026 (UTC)
The tag is a maintenance tag for calling attention to a possible problem, not a scarlet letter of shame. If you can't identify any issues then it shouldn't be tagged. Is the markdown you're concerned with the additional linebreaks, or is there other markdown I'm missing? ScottishFinnishRadish (talk) 17:27, 17 July 2026 (UTC)
The markdown here is the [link text](link url) formatting, which was automatically caught by an edit filter. (Before anyone starts to wikilawyer over this, the edit filter does not do anything to article text.)
As far as the "issues," the issue is that AI-generated text is not allowed on Wikipedia. As with copyvios, the guideline isn't "AI-generated text isn't allowed unless someone thinks it's OK," the guideline is that it isn't allowed at all. Thus the two solutions to the issue are to either rewrite it without AI or to remove it. Gnomingstuff (talk) 17:30, 17 July 2026 (UTC)
FWIW that was the intent/spirit of NOLLM when it was written, but someone other than the author would need to go through the article (Regulation to Prevent and Combat Child Sexual Abuse, why on earth would anyone use LLMs on such a sensitive topic) to ensure WP:V is met, which is a massive task (I guess spotchecks may suffice?). I think presumptive tagging is fine until someone does that. Deedubaroo shouldn't use LLMs in future because of this waste of community time; if it turns out they do regularly use LLMs responsibly, other than me eating my hat, they or you can propose user right for LLM use. We don't automatically assume or take someone's word that they use LLMs responsibly for very good reason Kowal2701 (talk, contribs) 17:56, 17 July 2026 (UTC)
The tagging wouldn't be presumptive though, since we don't need to presume that there is LLM-generated text here. It is an accurate reflection of the state of the article. ‑‑gurkubondinn 18:18, 17 July 2026 (UTC)
True, but this sort of discussion was what I intended to have at a later stage (ie. just get NOLLM passed as focussing on content quality for the time being), with a later RfC to remove For this reason which'd basically make it that LLM-generated content is prohibited on principle. The idea was that we'd have more time to have community discussions and see how things develop, rather than just internecine conflict and limbo. Imo we're not really at the point where such a RfC is necessary since this is so incredibly rare/uncommon, and if the article is otherwise PAG-compliant we should probably focus on lower-hanging fruit re cleanup anyway Kowal2701 (talk, contribs) 18:32, 17 July 2026 (UTC)
I have re-added the tag to the affected section for the sake of more important hills to die on, but I am at a loss as to how to progress on completing the remainder of the section, which was ideally going to be this weekend's project. If I add more text, it increases the size of the future independent review needed.
If you have influence on the course of the related policy (I am a baby Wikipedian), I think an approach similar to modern tech usage is fine: human review of every word (or as it goes in tech, change review by another human). For my purposes, the current state of affairs is not much far off banning typewriters, although I do understand its intent. Deedubaroo (talk) 18:53, 17 July 2026 (UTC)
One of the big reasons why WP:NOLLM is the way it is today is because anyone can generate hundreds of articles worth of unverifiable, undue, unencyclopedic content every single day. And currently people do do that every single day, and they have been doing it every single day for nearly 4 years, and there are only like five editors total who actually identify, verify, and cleanup LLM-generated content on a day-to-day basis, so the project of catching & reviewing it all is literally insurmountable, even now that editors can be blocked for persistently introducing LLM-generated text (a policy only in place since like March of this year). So if you're wondering why people here get this intense about LLM usage, that's why.
I'm aware this debt exists in tech too, where code reviewers are buckling under the weight of an overwhelming amount of LLM-generated code. I've heard of some open-source projects have responded to this with a policy similar to Wikipedia's, judging that not wasting volunteer time reviewing the 99% of crappy AI-generated security reports & pull requests is worth losing the 1% of good ones. purringmaggot (talk) 20:05, 17 July 2026 (UTC)
Working on a case here, I did a detailed source-to-text review of the first half of a medium-long article. It took me a bit over two hours, so about four hours might have done the whole thing. I did not fix it, mind you: just found the discrepancies. And I only did the English language sources, not the other three languages involved, because I don't feel I can do a good source-to-text in a language I don't speak.
We can't afford to spend 4+ hours on one article when the most prolific LLM users can introduce a long article in approximately 4 minutes.
So, backing up purring maggot: we are hard-ass because we have to be. M kuhner (talk) 21:23, 17 July 2026 (UTC)
This user keeps disruptively removing tags with snarky edit summaries. They removed only some of the LLM-generated text that they have added (seemingly only to remove the tag that they reluctantly re-added following this thread), but most of the text that they have inserted still remains in the article. Even text from edits like Diff/1364569730 remains in the article, and attempts to conservatively tag just that section resulted in the tag being disruptively removed in Diff/1364753188 (even though they have realised that AI generated content is exposed to the user). ‑‑gurkubondinn 09:44, 18 July 2026 (UTC)
The parent commenter is interfering with cleanup and acting without consensus that was established in this conversation. I believe they may be misusing the NOLLM rule to intentionally hinder further development of the content of the page owing to its topic. I'd like to document a progression:
1. They tagged the page originally. I responded with an immediate acknowledgement of the problem, indicated my unawareness of WP:NOLLM, and immediately indicated my willingness to comply with the rule going forward.
2. On requesting help with a procedure to remove the tag, they responded mechanically that all affected sections would require being rewritten, even though this is neither the text nor the spirit of the underlying policy.
3. On pushing back, they immediately escalated the issue, distracting from any cleanup happening yesterday owing to the conversation here, and without willingness to engage with any of the reasoning I offered for the pushback.
4. Following the discussion above, I wrote that I "re-added the tag to the affected section for the sake of more important hills to die on, but I am at a loss as to how to progress on completing the remainder of the section, which was ideally going to be this weekend's project".
5. In the interim, I realised the ideal approach was to inline-tag each affected paragraph, and work to verify, rewrite, and cite them one at a time. This is clear in the edit log: "tag every paragraph with the inline variant tag to facilitate one-at-a-time cleanup", and in line with prior strongly defended comments appearing on this thread, that the necessity of the tag is to reveal the presence of AI-generated text to the user.
6. This morning, less than a minute after submitting an obvious rewrite to the first section following this strategy, they reappeared making 3 edits in quick succession, re-adding the block-level tags in the wrong place, in spite of the fact every affected paragraph had already been tagged.
7. They then opened a talk page entry making a fresh accusation that the lead, an extremely carefully balanced piece of text on a controversial topic, was also AI-generated. It is not mentioned in the edit log, but I have been measuring exact word counts in this section, to ensure the disputed elements of the legislation are fairly represented. There is zero evidence for the accusation, and over a week worth of edit log refuting it. I simply do not understand their justification for claiming the lead itself is somehow AI-generated, absolutely nothing about the structure of the text whatsoever suggests AI use, it is simply too concise.
8. I reverted the block-level tag as it was clearly placed on the wrong section, it should, if it appeared anywhere, be at the top of the Provisions section, but a) every paragraph of that section is already tagged with a duplicate tag, and b) many paragraphs (both before the start of the incident and especially since today) are hand authored. A block-level tag is simply incorrect.
9. They have returned to this talk page to escalate again, telling more stories, and creating more distraction.
I am at a loss to explain the contributor's continual escalatory behaviour in spite of a cooperative author, baseless accusations, or their interest in this particular page. They originally stated they were "glad" to see my acknowledging the LLM issue openly, because such is rare. Yet, they continue to police the page excessively and without any level of ascertainable analytical effort, in a manner that is distracting from cleanup, and that also cannot be explained by my immediate willingness to cooperate with the policy.
None of the edits to this page from this user promote the goal of the encyclopedia or edits content policy, nor can they adequately be explained by a desire to even implement the WP:NOLLM policy, as all the affected content is already tagged, honestly and more broadly than the original tagged diff required for.
I am at a loss to explain continued interference in the progress of this page except out of the belief that as a tech worker, it is possible (or even likely) a political motive may exist. My only interest in editing the page was that it was in such a foul state originally, repeating almost exclusively partisan and often plainly incorrect claims and citations sourced from the Pirate Party, and documenting almost none of the non-controversial aspects of CSAR, that I took up the mantle to work on it as my first ever WP project.
My only request at this point is to leave the page alone if you are not actively intending to document CSAR. Working through the subtleties of this law is hard enough as it is without edit spam interrupting every change. The policing work is done - the policy violation was caught, acknowledged, and a plan to amend the affected sections is already in progress. There is nothing more to do on this page except actually document the subject in line with established policy and mission, which, as I feel it is necessary to remind the reader, is the collaborative authoring of an encyclopedia, not the implementation of a mechanical bureaucracy existing only for the purposes of practising the text of its rules free of any other productive output. Deedubaroo (talk) 10:40, 18 July 2026 (UTC)
Don't remove maintenance tags until the work is done. Not "a plan is in progress." M kuhner (talk) 12:37, 18 July 2026 (UTC)
The tags weren't removed, they were placed inline to facilitate the work. This must be what it feels like to attend a Wikipedia conference. I've readded the block tag, however utterly pointless that is, as it strictly duplicates the information from the inline tags, which I am actively using to clean up the page, while also suggesting non-AI text is in fact AI text. Thank you for helping improve the clarity and correctness of Wikipedia in this manner. If you'd like, I could also have ChatGPT whip up a fullscreen WebGL presentation on the merits of WP:NOLLM to insert into the lead. Deedubaroo (talk) 12:50, 18 July 2026 (UTC)
Template:In use is more or less what you're looking for here. Gnomingstuff (talk) 14:55, 18 July 2026 (UTC)
The relevant sections have now been gutted and stripped bare. I have removed the maintenance tag as it is no longer relevant. In particular, that means all nuance on the difference between Commission, Parliament, and Council positions has been lost. The idea that this in any way promotes neutrality or the overall mission of the encyclopedia is laughable. I'd like to repeat something said in the earlier talk section:

I appreciate the spirit of the tagging work done here, and the sizeable scope of the problem for any maintainer taking up the mantle, but it's important to notice, and I strongly encourage you not to ignore it, that blanket tagging without reasonable application is entirely equivalent (i.e. just as bad) as blanket LLM use.

I believe this very much encapsulates what has happened here. Blind application of policy is no better a poison than the ill it was meant to cure, i.e. this noticeboard is living, breathing and wholly intentional invocation of Goodhart's Law. That there is even a focal point of discussion for this policy can only act like a magnet of the worst kind. I hope that to at least someone reading this section, the point will ring home and hopefully moderate future applications of the policy.
In the meantime, I have a giant section to rewrite from scratch. Farewell, hope we don't cross paths again. Deedubaroo (talk) 15:47, 18 July 2026 (UTC)
@Deedubaroo, you quoted above a sentence and said it was spoken in the talk section above. No such thing was said, where did you get that from? Hooples (talk) 11:17, 19 July 2026 (UTC)
@Hooples, it seems like it was in § Oh darn, you're right of course on my talk page (where they also said that they believe that tagging articles with LLM-generated text as having LLM-generated text is "without regard for the spirit of the underlying policy"). ‑‑gurkubondinn 11:33, 19 July 2026 (UTC)
the seemingly unknowing of what page is being posted on and the frequent policy citing and poor wikilawyering is common on claude models atm. stricken due to it being too close to aspersions for my liking and not wanting to cause any upset Hooples (talk) 11:57, 19 July 2026 (UTC)
FYI: https://vocaroo.com/1i29z1KStfmT Deedubaroo (talk) 12:13, 19 July 2026 (UTC)
No clue why this is an audio file (never seen that happen before here), onwiki communication should stay onwiki. Externally hosted replies can disappear, be taken down, remove or otherwise altered. A quick transcription (made by Google Gemini), for others that don't want to listen to it:

I figured it would be easier just to send a voice note. I am, in fact, a real person. It is deeply demeaning and offensive to keep on suggesting that, for whatever reason, I'm using LLMs to write simple comments when the content of the comment should be very clear.

I'm actually really embarrassed that I've even contributed to Wikipedia after I discovered how it works internally. It's—it's absolutely disgraceful. The AI noticeboard, what are you people even doing? Like really, what are you even doing? You've—you know, you know you're worried about like machines taking over writing mechanical text, but meanwhile, every way you're acting is entirely mechanical. I've flagged, from the first post to the first talk page, the fact that the way you're acting doesn't actually sit with Wikipedia policy. You're enacting like machine-generated bloody rules. And every time something is said, it falls back to some ad hominem comment like, 'The seemingly unknowing of what page is being posted on and the frequent policy citing and poor Wiki-lawyering is common on Claude models.'

That's just plainly offensive. There's no other way to put it. I wouldn't like to work with you for any reason. I don't know what you think you're doing to help Wikipedia. Um, it's a really bad look. Um, I've been tempted to blog about it, actually—I'm not going to—but I have been tempted. Um, yeah, it—it's really disheartening to see this on a supposedly collaborative project. You know my account is 19 years old. You can see the edit log for yourself, and this is still the lazy-minder response. There's no place for this. There really isn't. It's just embarrassing.

As I said before, take care. I hope I don't bump into you again.
— User:Deedubaroo's audio file, transcribed by Google Gemini

I've listend to the audio file, and this is more or less accurate. ‑‑gurkubondinn 13:07, 19 July 2026 (UTC)
FWIW I've submitted an account vanishing request, this episode has been thoroughly persuasive regarding the value of further editing efforts, as they will inevitably provoke yet similar manifestations of some other sort while the actual work of documenting some sensitive topic (in this case CSAR) remains undone. My manual labour is better spent elsewhere, and that is an entirely natural consequence of how WP:NOLLM has been enacted here. To be clear, I believe this is an institutional problem manifesting through WP:NOLLM, not a problem of the policy itself. Deedubaroo (talk) 13:18, 19 July 2026 (UTC)
Just be cognisant of WP:VANISHNOT, if your vanishing request is granted. ‑‑gurkubondinn 13:26, 19 July 2026 (UTC)
renamed to Renamed user 1b81409746c7324294ce6693d0de173a (talk · contribs) Kowal2701 (talk, contribs) 09:06, 20 July 2026 (UTC)
id like to say, I withdraw any aspersions that I might have implied by my comment, as I DO believe you are a real person. This is the first time I've seen a voice memo sent as a wiki reply, made me raise my eyebrows for a moment. I do have one comment though, you're saying basically that people are too hard-lined about these AI rules yeah? It's been explained multiple times in this tall thread how much work it is to fix editors who use LLMs, especially ones who seem to have worse motivations than you. I appreciate you working on that article and I hope you understand that it's not a personal value judgement on you when people ask questions about possible LLM use. Hooples (talk) 13:12, 19 July 2026 (UTC)
I mean, it's not even only that. LLMs are simply incapable of producing text that is up to standards, and I offloads the work of verifying sources to other editors (while the user who inserted the references is off generating 5 more articles, requiring 5x more editors to verify those, and so on and so on). And just how using llm-generated text is just fundamentally contrary to what Wikipedia is. There is already at least one llm-generated encyclopaedia, and there's a reason that nobody uses it. ‑‑gurkubondinn 13:22, 19 July 2026 (UTC)
The user has now stated their intent to WP:VANISH and not edit Wikipedia again. Then the question is just what to do with the Regulation to Prevent and Combat Child Sexual Abuse article.
It was moved from Chat control Chat Control in revision 1363654197 on 11 July 2026. This was an undiscussed move, but nobody objected either. From a WP:COMMONNAME standpoint, it should maybe be moved back though. In revision 1363793750 from 12 July, the page was very clearly LLM-generated in large parts. Some users, like JSwift49, seem to have done a fair job at trying to fix various issues, but I'm not sure how much of that remains. ‑‑gurkubondinn 13:51, 19 July 2026 (UTC)
I wouldn't be against moving it back to Chat control, although I think more discussion would be useful since it is a term primarily used by critics. JSwift49 15:20, 19 July 2026 (UTC)
Minor note, but it was Chat Control in title case. ‑‑gurkubondinn 20:05, 19 July 2026 (UTC)

Kned Wiki

Latest comment: 1 month ago2 comments2 people in discussion

Renamed user 77899991399dbbe0009e5042335b619b (talk · contribs), XTools

Good-faith editor, but admitted to using AI, and vanished after being informed of NOLLM . 123 creations, the first of which was (Nov 2024) which has WP:SUPERFICIAL and WP:AIVOCAB eg. ...the gardens feature panoramic views over Western Sicily and hold cultural and historical significance. The Balio Gardens are part of the Grandi Giardini Italiani network, which showcases Italy’s prominent historic gardens., and WP:AILIST, WP:AITITLECASE etc. Analysis needed of their later creations Kowal2701 (talk, contribs) 17:00, 18 July 2026 (UTC)

First warned about LLM use in § May 2025, and then admitted to LLM use in § May 2026 (saying I can 100% vouch for the content) after which they came to my talk page in § Trapani-Berghi to argue about their LLM-generated edit. After I actually relented and reinstated it, I had to then tell them in § Removal of maintence tag to not remove the tag and also told them about NOLLM at the same time. That doesn't seem to have encouraged them to stop, because I seem to have noticed them again and when I asked § Are you still using AI to edit?, after which they promptly blanked their talk page again and vanished.
I think that basically every edit and page created by this user can be presumed to have been LLM generated. They seemed like a very heavy user. ‑‑gurkubondinn 19:44, 18 July 2026 (UTC)

User:Nish Aryal

Latest comment: 22 days ago8 comments5 people in discussion
NO ACTION
No consensus on LLM use. (non-admin closure) M kuhner (talk) 23:18, 10 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Red X symbolN Cleanup is unnecessary.

Though they've said otherwise here, Nish Aryal (talk) seems to be using LLMs in some of their contributions, specifically in all of their edits marked with the tag "Edit Check (paste) shown." See this diff at Dal bhat for example: . All their other edits with that tag show similar LLM-like language. InfernoHues (talk) 16:59, 23 June 2026 (UTC)

They've made a few drafts but I'm not seeing any really blatant evidence, might be one to keep an eye on? Kowal2701 (talk, contribs) 18:03, 24 June 2026 (UTC)
"Soupy lentil soup" is a weird phrase but I don't see much that's obviously AI. Looks more like stilted non-native English prose, possibly using a thesaurus. Some of the copy and paste warnings are on single phrases. SenshiSun (talk) 04:42, 26 June 2026 (UTC)
Now at ANI, WP:ANI#Repeated ECR vios in the WP:CT/CASTE space Kowal2701 (talk, contribs) 06:08, 22 July 2026 (UTC)
Does not read like AI - reads like someone who isn't the most fluent in English Jishara (please ping upon response) (talk) 01:51, 4 August 2026 (UTC)

I'm not seeing a consensus for cleanup here. Are there any articles where we feel sure that LLM was used? Otherwise this maybe should be closed as inconclusive. M kuhner (talk) 01:22, 10 August 2026 (UTC)

I'm fine with that. I don't think they've done that much mainspace writing either way. It was mainly their edit requests that concerned me. InfernoHues (talk) 01:40, 10 August 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Another AI agent blocked

Latest comment: 22 days ago31 comments6 people in discussion
EDITOR AND SOCKS BLOCKED
Cleanup completed. (non-admin closure) M kuhner (talk) 14:43, 10 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Green checkmarkY Cleanup has been completed, see /2026-07-12 Jaredzimmerman for reference.
LLMPRODs are done

User:IndigoKarasu was blocked by User:Asilvering for being an AI agent. Do we need to get formal consensus for WP:LLMPROD? Helpful Raccoon (talk) 04:22, 30 June 2026 (UTC)

First, I think we'll want to see what the editor (or their AI agent) has to say for themselves. In solidarity, asilvering (talk) 04:55, 30 June 2026 (UTC)
I fixed the Frank Epperson article. EaglesFan37 (talk) 02:22, 7 July 2026 (UTC)

I think we have now seen that. In my opinion the unblock request is LLM. I'd ask that any cleanup focus on the declined drafts rather than articles the community already accepted. just sounds too much like all the other requests to "focus" on something other than doing a full investigation. In any case it was declined. M kuhner (talk) 21:15, 8 July 2026 (UTC)

Indeed, I don't think we're going to get anything useful out of this person. In solidarity, asilvering (talk) 03:21, 9 July 2026 (UTC)

Support immediate cleanup. M kuhner (talk) 03:58, 9 July 2026 (UTC)

Updated, their LLM use began 22 March 2023. Some details found: WP:OAICITE , imparted a foundational understanding of blending historical styles with modern sensibilities , a notable edifice that exemplifies a meld of Brutalist aesthetics with the tropical ambiance of Hawaii , various 404s in .
When compiling the list I also found evidence of programmatic TA socking to "fix" pages they created, see the SPI for details. fifteen thousand two hundred twenty four (talk) 19:47, 25 July 2026 (UTC)
Wow. Edits to one article by 14 different TAs, all with very similar sounding edit summaries, between 13:12 and 18:03 on the same day. (And then IndigoKarusa started in.) I'd be interested in a community view on whether articles of this sort (this one is Tropical Modernism) can be LLMPRODded on the grounds that this is not really "substantial input from other editors". A subsequent edit summary by Gnomingstuff indicated that they were not satisfied with the cleanup. M kuhner (talk) 17:40, 28 July 2026 (UTC)
I'd wait for the SPI to resolve before prodding to prevent any confusion. fifteen thousand two hundred twenty four (talk) 17:43, 28 July 2026 (UTC)
Sounds good. I take it that's an implicit "Yes, you can LLMPROD despite substantial contribution from 'other' editors"? M kuhner (talk) 17:47, 28 July 2026 (UTC)
If those other accounts are clearly socks, then yes, but what is clear to someone with deeper knowledge of a case may not be clear to an admin responding to an llmprod, thus waiting for SPI to make blocks. fifteen thousand two hundred twenty four (talk) 18:07, 28 July 2026 (UTC)
Blocked. Go nuts. In solidarity, asilvering (talk) 02:10, 29 July 2026 (UTC)

I am informed that edits by PigeonChickenFish need not be considered in determining whether articles for this case have significant input from uninvolved editors, and I am proceeding accordingly. (A good thing, too, or nothing could be LLMPRODed.) M kuhner (talk) 17:52, 29 July 2026 (UTC)

For the record: PigeonChickenFish has a connection with Jaredzimmerman and also abused their reviewer permission to support their edits . fifteen thousand two hundred twenty four (talk) 18:21, 29 July 2026 (UTC)
I LLMPRODed a few more on that basis. There are still a few that I could only tag and hope. M kuhner (talk) 19:27, 29 July 2026 (UTC)
If someone wanted to look at Ernest_Hideo_Hara I'd be grateful. Got the rest of the Jaredzimmerman, starting on the others. M kuhner (talk) 19:32, 29 July 2026 (UTC)
Prodded, creating IP is clearly them based on timing, edit summary use ( vs and vs ) and matching recent TA behavior detailed at the SPI (WP:TAIV not used). fifteen thousand two hundred twenty four (talk) 19:48, 29 July 2026 (UTC)

Done! Just need to watch till the LLMPRODs and one AfD go through. (Risk of sockpuppet removals seems fairly high.) M kuhner (talk) 19:51, 29 July 2026 (UTC)

I checked the TAs in the SPI report; none had edited any pages beyond what's in the tracker. M kuhner (talk) 19:57, 29 July 2026 (UTC)
Almost done here but there are a few dePRODs. Not worried about the one by Cunard, who has rewritten the entire article, but the others could use checking--claim is made that there was no LLM, not that cleanup was done, and I am skeptical of that. M kuhner (talk) 19:42, 6 August 2026 (UTC)
@Dan arndt, can you please clarify what you mean? Normally I'd restore the LLMPRODs because WP:LLMPRVOBJ wasn't followed (i.e. the sources weren't checked for WP:V), but they'd be eligible for deletion immediately and that felt too combative, so happy to discuss Kowal2701 (talk, contribs) 19:48, 6 August 2026 (UTC)
I looked at supposedly-cleaned Tropical Modernism and the first two cites I spotchecked, under Exemplary projects (which also has some botched cut-and-paste visible) did not validate. I will post to the dePRODer's Talk page. M kuhner (talk) 14:10, 8 August 2026 (UTC)
Restored prod, not only was that an improper removal, the text is also full of signs of model generated text. fifteen thousand two hundred twenty four (talk) 19:57, 8 August 2026 (UTC)
Could you restore the prod to Karunaratne House as well? I just did a detailed check and the source/text is not okay. I was the original LLMPRODer and feel uncomfortable about restoring against opposition. It's a shame, the idea of the article is interesting--but the text is LLM. M kuhner (talk) 23:10, 8 August 2026 (UTC)
I did a quick spot check and did not immediately see any V issues that would indicate LLMPRVOBJ wasn't followed, if you can give some specific examples I'll consider them. fifteen thousand two hundred twenty four (talk) 00:05, 9 August 2026 (UTC)
I am hampered by not having access to the key reference (1), but I found these:
The house was commissioned in 1947 by Algy and Letty Karunaratne, friends of de Silva's family sourced to (3), which says it was her father's business partner (not friend) Algernon (not Algy) and does not mention Letty at all.
The house is frequently cited in analyses of de Silva's work and of Tropical Modernism more broadly is sourced to 2 articles about de Silva, but I don't see how they can support "cited in analyses [...] of Tropical Modernism more broadly." And it's such a typical LLM sentence.
most of her other Kandy commissions have been demolished due to development pressure is sourced to (1) and (5). (5) says many of her creations have been demolished, but doesn't confirm "most", "Kandy", or "development pressure." I can't see (1) but if (1) has that, why cite (5)?
It's not terrible: if it weren't by the author of some pretty terrible LLM articles I would probably let it slide. Maybe someone else has access to (1)?
M kuhner (talk) 00:28, 9 August 2026 (UTC)
Maybe someone else has access to (1), archive.org has it . fifteen thousand two hundred twenty four (talk) 00:44, 9 August 2026 (UTC)
Thank you!
Finishing the analysis: "steep hillside overlooking Kandy Lake" is sourced to "high above Kandy Lake in the wooded hills" which is not a very close match.
Where (1) says "glazing" article has "large windows", and the claim this was to reduce heat is not in the source.
The biggest issue:
(1) says In 1953, MARG devoted 11 fully illustrated pages to the house. David Robson, the leading authority on the work of both Bawa and de Silva, notes that ‘it read like a manifesto for a “Modern Regional Architecture” and was very influential.’ This quote is attributed to MARG in the article: the journal MARG, which de Silva had co-founded in 1945—devoted an illustrated feature to the house, describing it as a manifesto for a "Modern Regional Architecture." But while I can't find a birth date for Robson, he was writing books on Bawa in 2007, 54 years later. I don't think he said this in MARG (also (1) uses "notes" as if Robson's statement was contemporary with (1), not the past tense they use for MARG). I think Robson said this about the MARG article at a considerably later date. Furthermore, the next line in the Wikipedia article is The architectural historian David Robson identified the feature as influential in shaping discourse around regional modernism in South Asia. also sourced to (1), and this goes far beyond what Robson said--plus, the same quote has now been described twice, once attributed to MARG, once to Robson.
My best attempt at a search (though it may be contaminated by LLM) says Robson said this in 2015.
The "Algy and Lettie" that has been attributed to (3) is in (1) and not in (3).
I personally think that these are LLM sourcing issues. M kuhner (talk) 02:38, 9 August 2026 (UTC)
Upon closer review the Architectuul source is also user generated, and when considering this along with the issues you've raised it appears that the objection process wasn't followed, I've restored the tag. fifteen thousand two hundred twenty four (talk) 19:38, 9 August 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

User:Liamb723

Latest comment: 21 days ago8 comments3 people in discussion

This user has been pumping out long articles with 30-60 seconds between articles today, and is now removing {{AI-generated}} tags from the articles. ‑‑gurkubondinn 23:00, 24 July 2026 (UTC)

Hi, I understand your concerns, but on my userpage under Future Projects, you can find sandbox articles of every single article, and if you check the edit history of these sandbox pages, you can see that every page actually cost quite some time to write. After finishing all sandbox pages, I copied the source to a new article. I did not use generative AI anywhere in the articles. Liamb723 (talk) 05:14, 25 July 2026 (UTC)
An example: https://en.wikipedia.org/w/index.php?title=User:Liamb723/sandbox&action=history Liamb723 (talk) 05:44, 25 July 2026 (UTC)
You're still asking me to believe that Diff/1364253988 was written in under two minutes, so I find the claim that you did not use generative AI very hard to believe. ‑‑gurkubondinn 15:38, 25 July 2026 (UTC)
I am home in a few hours, I will get back to you Liamb723 (talk) 16:13, 25 July 2026 (UTC)
The edit you mention @Gurkubondinn was a copy of the sandbox page Po-Eun (teul). I will try to explain everything in more detail and with more proof.
A month ago, I came up with the idea to create articles for each Taekwon-Do form (teul) of the style I practise and one of the most practised TKD styles worldwide. There are dozens of these patterns, so I knew it would be a lot of work. I started with the first pattern in a sandbox page, and when I liked the look of that, I copied the source to the sandbox page of the second pattern, and then to the third, et cetera. To prove this, I would like to give an example: when creating the sandbox page for the 22nd pattern, Seo-San, I copied the entire page of the sandbox article of the previous pattern. As you can see, the contents coincide completely (except for some later improvements I added to the sandbox article on Mun-Mu. Another example is here: the first edition of the sandbox page of Yeon-Gae was the same as the preceding sandbox article Choe-Yeong.
Why did I do this? By copying the source of an article about a pattern to the next one, 1) I was sure that the structure of all articles was similar, and I find consistency in a set of articles about different elements of the same subset important, as well as that this way I was sure I would not forget something, and 2) it saved me a lot of time, because I did not have to rewrite the entire article, but could instead reuse the same sources when relevant and change information and data that is different across the patterns. This way, it sometimes only took me 30-45 minutes to create a new article, as I could use the structure of the previous article and had all sources by hand.
So how did the process look like? 1) I created the first sandbox page from scratch and started expanding it, 2) I copied what I had to the next article, and then changed the title, information, data, etc, but maintained the structure, 3) I kept copying and editing this way until finishing the list (while improving the previous pages if I thought of a relevant new piece of information on a page), 4) and then copied the sandbox articles of all 29 pages to actual Wikipedia entries. I believed this was a safe way to take my time and be able to work on all 29 pages calmly before publishing them.
In hindsight, I can completely imagine how this looks like a series of AI-generated pages, and I definitely should not have removed the AI templates myself, I simply believed it was a false automatic reaction by Wikipedia after publishing many pages at the same time. I would like to apologise for this removal. The articles are not written, partially or completely, by generative AI. If you have any remaining doubts, please feel free to go through the edit history of all sandbox pages I mentioned, as they are all listed on my Userpage as a sort of personal to do list. If you have any questions, please feel free to ask them. Liamb723 (talk) 18:17, 25 July 2026 (UTC)
(sorry for the late response, this page is very busy) @Liamb723, I believe you, at a close look the articles don't read as AI, and I put a few into Pangram and they came back as 100% human-written. Sorry about this, and thank you for your honesty, you're doing great work. Kowal2701 (talk, contribs) 12:06, 11 August 2026 (UTC)
Thank you! Liamb723 (talk) 12:15, 11 August 2026 (UTC)

Logistics in South Africa

Latest comment: 22 days ago13 comments8 people in discussion
NO ACTION TAKEN
No consensus of LLM use. (non-admin closure) M kuhner (talk) 21:30, 10 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

I believe the page Logistics in South Africa is written by LLM. It was imported as a 40k bit file and its sentence structure uses lots of (abrv.) abbreviations in parenthesis. The user is creating a lot of pages, mostly about African companies.~2026-42422-32 (talk) 14:02, 30 July 2026 (UTC)

I can't say any of my usual checks presented anything. Indeed referencing "Joburg" is something that local residents would tend to do automatically, and there is some also very SA style grammar and tone here (dropping the subject, abbreviating text before the verb). Not so sure about some of the sources though, in terms of independent reliable sources. ChrysGalley (talk) 14:27, 30 July 2026 (UTC)
Something is shady, the editor created 109 pages in 5,000 edits and they are mostly about African companies, then the rest of the content is about industry in Africa. Check this link
https://xtools.wmcloud.org/pages/en.wikipedia.org/Husskeyy ~2026-42422-32 (talk) 16:23, 30 July 2026 (UTC)
I think we better hear from the editor concerned. I don't think LLM is in the mix, but there are other concerns here. ChrysGalley (talk) 16:39, 30 July 2026 (UTC)
Something is shady, the editor created 109 pages in 5,000 edits and they are mostly about African companies - So essentially 2% of their edits are article creations? That does not sound suspicious at all to me; many prolific human article writers have a similar or higher ratio of article creations to total edits. Prolific human writers tend to specialize in specific topic areas. Now, I haven't checked whether the articles comply with policy, but this does not ring alarm bells AI-wise. Accessedgrant (Epicgenius mobile alt) (talk) 05:13, 31 July 2026 (UTC)
Husskeyy (talk · contribs), XTools Kowal2701 (talk, contribs) 14:32, 30 July 2026 (UTC)
A cursory examination of editor behaviour reveals that it is in my opinion inconsistent with LLM use and the page is absent most modern LLM tells. Indeed LLMs tend not to use abbreviations in my experience. Fermiboson (talk) 15:06, 30 July 2026 (UTC)
LLM’s constantly use “World Trade Organization (WTO)” which we don’t use on Wikipedia. ~2026-42422-32 (talk) 16:30, 30 July 2026 (UTC)
There are only 41 hits for that , most of the ones I looked at were added pre-Nov 2022 Kowal2701 (talk, contribs) 16:40, 30 July 2026 (UTC)
Regarding "which we don't use on Wikipedia", I'm not sure that's correct. It's consistent with the guidance given in MOS:ACRO1STUSE. And if a user is repeating "World Trade Organization (WTO)" after the first use, I think that would be more of an indicator of human writing, since this is what people writing promotionally sometimes do. Accessedgrant (Epicgenius mobile alt) (talk) 05:04, 31 July 2026 (UTC)
Nothing really stands out in either direction as far as I can tell Gnomingstuff (talk) 15:22, 30 July 2026 (UTC)
Definitely does not read as LLM to me - no LLM tells that I can find. Could be promotional or CoE, or could be an editor who just really cares about this stuff. IMO, definitely not LLM. Jishara (talk) 01:31, 1 August 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

User:Panecashio

Latest comment: 26 days ago5 comments4 people in discussion
mainblock  Panecashio (talk · contribs · logs · block log) (6 August 2026)
Editor has been blocked from editing the mainspace (articles only) for AI/LLM violations. 
Panecashio (talk · contribs) Kowal2701 (talk, contribs) 22:09, 5 August 2026 (UTC)

User joined 8 days ago and has created multiple articles, mostly biographies, since, all very similar to each other. Several edit summaries ( ) are clearly AI-generated. Most other edit summaries use the same canned summary of “Creating draft for [subject] using reliable source references.” or say “Fixed Error” regardless of what was actually done. All articles created by user are started as one large edit. Smaller edits by the user () display a very poor grasp of English inconsistent with the larger edits. Several articles () declined for LLM use, which the user ignored and moved into the mainspace anyway. User was asked about possible LLM use and did not respond.

MCRPY22 (talk) 19:01, 5 August 2026 (UTC)

@Admins willing to patrol AINB: Request a block from mainspace to get this editor to communicate. Their Talk page is littered with concerns about LLM use, copyvio, and non-notability. They have denied LLM use but I don't find this credible. They don't otherwise respond to concerns. M kuhner (talk) 01:45, 6 August 2026 (UTC)

Red X Blocked Panecashio from article space for LLM misuse. See nonexistent categories in .​ — Newslinger talk 02:33, 6 August 2026 (UTC)
They've had their block appeal denied but have agreed to work in draftspace and go through AfC. I'm gonna draftify and tag their creations Kowal2701 (talk, contribs) 15:25, 6 August 2026 (UTC)

User:Edfl22

Latest comment: 1 month ago4 comments3 people in discussion

Edfl22 (talk · contribs · deleted contribs · logs · filter log · block user · block log) is the smoking gun. User has remained active with a varying number of non-obvious signs and very seldom are their edits long enough to tell in isolation but are all individually... weird. The most useful one is which I can only characterise as "compulsively editorialising". Fermiboson (talk) 10:15, 12 July 2026 (UTC)

Another smoking gun: AbuseLog/42097809
‑‑gurkubondinn 11:27, 12 July 2026 (UTC)
I will preface that the manner of editting (deemed as "compulsively) was not to alter the plot but rather to make the plot clear. It wouldn't make sure to condense the article. In addition, I actually watched the film last night so I figured a few details had to be expanded. In addition, the film has a low to zero sources regarding the film's breakdown, other than the film itself. Will the article be reverted to it's original version? Edfl22 (talk) 17:40, 12 July 2026 (UTC)
That depends on the degree to which you used an LLM, which it would behoove you to be transparent about. Fermiboson (talk) 17:52, 12 July 2026 (UTC)

User:ReaderOfSciFiNovelsNPhilosophy

Latest comment: 23 days ago14 comments6 people in discussion

There is a discussion with the user at User talk:ReaderOfSciFiNovelsNPhilosophy § June 2026.

This is an OKA translator, has twice inserted OAICITE into articles. Has been making large edits that will be difficult to check. ‑‑gurkubondinn 13:33, 16 June 2026 (UTC)

Another OKA translator? Keep seeing them everywhere Fortek67 (talk) 16:41, 16 June 2026 (UTC)
Remind me again why we allowed them... did people actually think that anything good could come of the combination of "AI" and "editing for pay"? Stepwise Continuous Dysfunction (talk) 19:21, 16 June 2026 (UTC)
IMO it's not a bad idea but users should be experienced Wikipedia editors. I've previously rejected AfC submissions from these editors, because after heavily reviewing them, these articles are poorly reviewed by the editors. I think OKA editors should be fluent in both English and the translating language, be familiar with the topic, and properly review these articles. Fortek67 (talk) 19:29, 16 June 2026 (UTC)
Looking at Special:Contributions/7804j does not exactly inspire confidence either. Created 19 new articles two days ago, sometimes as little as two minutes apart. ‑‑gurkubondinn 09:07, 17 June 2026 (UTC)
Yesterday they created 10 articles in the span of about 1 hour, with as little as 3-4 minutes between articles. They have yet to answer my question of how they propose validating 162 sources added in one paid edit, and have still not identified the OAICITEs inserted with those paid edits. How is this supposed to be sustainable? ‑‑gurkubondinn 08:47, 22 June 2026 (UTC)
And yesterday they created five articles, four of those were created a the span of 30 minutes. ‑‑gurkubondinn 21:29, 26 June 2026 (UTC)
the perverse incentive strikes again Gnomingstuff (talk) 22:53, 16 June 2026 (UTC)
Tried to point that out in User talk:ReaderOfSciFiNovelsNPhilosophy § Disambiguation link notification for June 16, but no response yet. I think it's a bit much to pay someone to add 162 sources that they (presumably, by looking at the timestamps) haven't read and then expecting other people to read and verify them for free. ‑‑gurkubondinn 09:04, 17 June 2026 (UTC)
Don't ask me, I agreed with the suggestion that OKA editors would be "prohibited from directly editing articles entirely" (in the capacity of inserting paid machine translations on behalf of OKA). ‑‑gurkubondinn 16:48, 17 June 2026 (UTC)

I am wondering if this has to be escalated (AN? ANI?) as Gurkubondinn's conversation with the OKA head did not go anywhere useful. This editor seems to have stopped editing, but if the project head cannot accept responsibility for the people they are paying to edit Wikipedia, that's a serious problem. M kuhner (talk) 19:07, 8 July 2026 (UTC)

In case you haven't seen the previous community discussion about OKA: Wikipedia:Administrators' noticeboard/Archive378 § OKA: problematic paid translation and lead rewrites via LLMs across thousands of articles.
Which lead to these editing restrictions for all OKA volunteers: WP:EDR § OKA
‑‑gurkubondinn 19:26, 8 July 2026 (UTC)

This user is now active again, and summaries like Diff/1366326652 suggest that they may be trying to fix LLM-generated text now. ‑‑gurkubondinn 17:17, 28 July 2026 (UTC)

Yep, they're still doing rewrites, should get a barnstar once they're done Kowal2701 (talk, contribs) 16:57, 9 August 2026 (UTC)

User:Glm1

Latest comment: 1 month ago6 comments2 people in discussion

Glm1 (talk · contribs)

Account created Feb 2025, ~2000 edits (mostly minor or content removal), no article creations. LLM-generated ANI report where they denied using AI (July 2025), edit with markup (May 2025), last edited 27 June 2026. I've asked them on their talk. Would appreciate if someone could analyse their recent edits. Kowal2701 (talk, contribs) 16:41, 12 July 2026 (UTC)

Looking at Larry Gatlin discography June 27 edits.
First they added an Adult Contemporary column in two edits. Most of an hour later, they came back and added two cites to Billboard, but did not know how to make them into inline citations--put then in the reflist, so they are sitting at the bottom of the article detached from what they were meant to support. Citations are rudimentary: you can find the article from that with a search and a lot of clicking, but why not link directly to the PDF? It is possible to work out which two songs are meant and compare with citations. One is entirely correct. In the other, a claim about the Adult Contemporary chart is being sourced to an Easy Listening chart. (Which might be correct, I dunno, maybe those are different historical iterations of the same chart.)
I am trying to imagine how a human would do this using Billboard. You want the top charting, not just any chart where the song happens to appear: so you'd have to click into and search multiple issues. It is agonizing (I just tried it, so I know). I can't really reach a conclusion from these edits but I'm leaving them here because they were so much work! M kuhner (talk) 21:54, 12 July 2026 (UTC)
Same type of chart gnoming in The Statler Brothers discography, same attempt to put the reference in the reflist, no URL despite online source. This time the intended location for the source was the newly added Contemporary column, which was never going to work because a separate cite would be needed for each entry (they are all in different issues of Billboard). I do not see any sign of LLM here, just someone very persistent (no one else could get any good out of those Billboard back issues) but with a poor grasp of referencing. Edit summary also confirms that (the editor believes, at least) Contemporary == Easy Listening.
Editor manages not to have learned how to do an inline citation in 2K edits. I will put a message on their Talk page and see what happens. M kuhner (talk) 23:33, 12 July 2026 (UTC)
I am looking at the edits to Buffy Sainte-Marie and they seem concerning. In particular, says in Wikivoice that Sainte-Marie was aware of the upcoming documentary when she retired. This is sourced to something that simply announces the retirement, and to . The latter source does repeat the allegation, but sources it to unnamed people connected with the documentary. I think this is too fragile to support a controversial statement in a BLP. But I don't know if it's a bad LLM paraphrase or something else --probably POV pushing, as seen in the change of the section title. (Someone else just reverted one of their edits on the basis of POV-pushing.) Still have no idea if there is LLM, though the edit summaries are suspicious. M kuhner (talk) 03:54, 13 July 2026 (UTC)
Update: they denied any LLM usage [1] Kowal2701 (talk, contribs) 22:08, 12 July 2026 (UTC)
[1] Kowal2701 (talk, contribs) 16:45, 16 July 2026 (UTC)

Nasvetmcbafee

Latest comment: 1 month ago3 comments2 people in discussion

Note: tracker includes Lanternspaper92 and Edit By Alex sockpuppets as per SPI report.

blocked sockpuppet in march, here LLM bolding, there ~2026-38678-64 (talk) 21:59, 13 July 2026 (UTC)

Their creations are eligible for WP:G5, but I'm going to WP:LLMPROD instead to give someone the chance to rewrite them Kowal2701 (talk, contribs) 10:49, 24 July 2026 (UTC)
FWIW, an admin at SPI also came to same conclusion, Wikipedia:Sockpuppet investigations/Lanternspaper92/Archive Kowal2701 (talk, contribs) 10:53, 24 July 2026 (UTC)

User:MaryTheo2012

Latest comment: 1 month ago4 comments4 people in discussion

This user was brought to my attention in § "Cleaning" on my talk page by 1000mm, who is an admin on nowiki and has blocked the user there.

The user seems to have been mostly concerned with inserting machine-translated versions of the MCND article to Wikipedia projects in various languages that they don't speak. So if anyone here is active on those wikis or speaks the language, you may want to help out over there (the page m:Artificial intelligence/Policies by project may be helpful to find the various guidelines and policies on LLM-generated content for other wikis). ‑‑gurkubondinn 10:26, 14 July 2026 (UTC)

same thing going on at the candy candy article ~2026-38678-64 (talk) 05:38, 15 July 2026 (UTC)
That is a wonderful page and maybe should be linked from the header of this noticeboard. M kuhner (talk) 21:15, 17 July 2026 (UTC)
Not about AI usage, but I've been repeatedly frustrated by this user's editing. They keep circumventing AFC without proving notability and using unreliable sources. grapesurgeon (talk) 23:15, 17 July 2026 (UTC)

User:Varshith.Reddy.90

Latest comment: 1 month ago5 comments3 people in discussion
mainblock  Varshith.Reddy.90 (talk · contribs · logs · block log) (16 July 2026)
Editor has been blocked from editing the mainspace (articles only) for AI/LLM violations. 

Varshith.Reddy.90 (talk · contribs · deleted contribs · blacklist hits · Edit filter search · what links to user page · count · COIBot · Spamcheck · blacklist hits · user page logs · x-wiki · status · Google · StopForumSpam) There are way too many pages for me to go through myself. It appears they're using an LLM to rewrite plot sections of articles. Jerod Lycett (talk) 00:47, 15 July 2026 (UTC)

This editor has 35 edits today, some of which are just 6-7 minutes apart despite being extensive rewrites/additions, and they have been (somewhat) warned about LLM usage already on the 10th but have never used their talk page before, so I'm hoping for a block from article space if they don't respond to your warning pretty soon. purringmaggot (talk) 02:11, 15 July 2026 (UTC)
Continued to do it. Gave a level 3 now. They're gonna get AIV'd soon from this. Jerod Lycett (talk) 06:42, 16 July 2026 (UTC)
@Admins willing to patrol AINB: Could we get a partial block so this editor might notice the warnings and agree to stop using LLMs? They have never used their talk page and continue to make apparently LLM-generated rewrites of plot summaries over and over again after 3 warnings. purringmaggot (talk) 18:22, 16 July 2026 (UTC)
Blocking here isn't a slam-dunk. Most edits are about plot summaries, which require no sourcing. I ran one extensive plot summary through GPTzero and it said "human written with AI polishing". In any case I have issued a mainspace block for AI usage and failing to collaborate (user has zero talk page edits). ~Anachronist (who / me) (talk) 18:42, 16 July 2026 (UTC)

User:Karaokelegend

Latest comment: 24 days ago2 comments2 people in discussion

Another large, high-volume, and ongoing spree of AI edits. They denied using AI but I think it's pretty unambiguous in this case; the smoking gun here is the [span_1](start_span) formatting in this edit, which upon googling seems to be a Gemini bug. (It also cites Grokipedia. What are we even doing here) Gnomingstuff (talk) 14:58, 17 July 2026 (UTC)

I am quite concerned at the false denial. Obvious AI usage in the linked diff. Binksternet (talk) 17:37, 8 August 2026 (UTC)

User:SilentHalo

Latest comment: 1 month ago5 comments4 people in discussion

Looks like SilentHalo is using AI to generate content, posts in ANI & AFD discussions, and reply's on talk pages. GPTZero comes back at 100% for most of them. They claim to be using Grammarly for copy editing but at the ANI discussion post it looks like there were several hallucinations. ~2026-40763-89 (talk) 19:22, 20 July 2026 (UTC)

Hi there. I replied to the user on their talk page. I'm not sure if using copy-editing tools is prohibited here. Wikipedia:Grammarly doesn't have much useful advice—it just says that it's prohibited for encyclopedia articles. SilentHalo (talk) 19:27, 20 July 2026 (UTC)
The relevant policies are WP:NOLLM and WP:LLMCHAT. LLM use, with extremely narrow exceptions, is not allowed in encyclopedia articles. Noticeboard comments that appear to be LLM can be collapsed and ignored.
The reason Grammarly is prohibited is that recent versions are, in fact, LLM, and can change the meaning of text substantially under cover of "improving" it. It can't be considered a "copy-editing tool" anymore.
Please completely stop use of LLM to generate text on Wikipedia. You might also consider helping with cleanup of your previous articles, which are otherwise likely to be deleted. This is a recoverable mistake--newcomers aren't necessarily aware of our policies and we're willing to work with you here--but only if you commit to stopping LLM use immediately. M kuhner (talk) 19:55, 20 July 2026 (UTC)
I didn't know that Grammarly was now an LLM. I'm sure it worked just like Microsoft Editor, but it was much better. I've always made sure that the tool hasn't made any significant changes. However, I'll abide by the policies. I promise never to use it again. SilentHalo (talk) 20:08, 20 July 2026 (UTC)
Not your fault, Grammarly is a little infamous for being misleading and sketchy about how its product works (for instance the whole debacle over a feature claiming to incorporate "expert feedback" by a bunch of writers who weren't involved in the feature or even told about it) Gnomingstuff (talk) 21:44, 20 July 2026 (UTC)

Urug1948

Latest comment: 1 month ago7 comments2 people in discussion

Urug1948 (talk · contribs) seems to use LLMs, i think that it might be the ghost of Calhetan77 ~2026-38678-64 (talk) 20:14, 20 July 2026 (UTC)

I've moved the SPI page between namespaces for you. ‑‑gurkubondinn 20:43, 20 July 2026 (UTC)
that broke my permlink! ~2026-38678-64 (talk) 21:04, 20 July 2026 (UTC)
Serves you right for refusing to create an account for no good reason![Humour] ‑‑gurkubondinn 21:17, 20 July 2026 (UTC)
it's a really good, compelling reason though! [humor] ~2026-38678-64 (talk) 21:29, 20 July 2026 (UTC)
humour (haha) aside, there's urug1948's SPI case (before i saw them as a likely sock of Calhetan77) which might be of good use to clean up ~2026-38678-64 (talk) 01:07, 21 July 2026 (UTC)
The user that filed that SPI was blocked as a WP:DUMBSOCK... ‑‑gurkubondinn 09:14, 21 July 2026 (UTC)

~2026-40064-42

Latest comment: 1 month ago8 comments3 people in discussion

This temp account claims they only used AI for copyediting but I know that new accounts don't make articles that good. I don't think the copyediting allowance should even be allowed. The source editor already has red underlines for typos. You don't need AI. The temp account already reverted me, so I don't want to edit war with them. ~WikiOriginal-9~ (talk) 12:39, 21 July 2026 (UTC)

"New accounts don't make articles that good" is not reliable: I got my first one accepted on the first try, no problems. We need a detailed analysis. I'm looking at material re-added in this: .
rated the No. 4 overall prospect by scout.com is sourced to KSL.com, which is not ideal--should go to the original--and it's odd that they don't cite the No. 1 from rivals.com, also mentioned there. (But I don't know football--maybe the #4 is more prestigious than the #1.)
I don't think the author of this text understood Aiono's college career when writing the early parts, as they are muddled; there is a source that lays it all out, but the edit reads like they had written most of it before finding that source.
Maybe a football fan could tell if "half a sack" in the article is a match to "assisted tackle for loss" in the source.
The source given for Aiono graduated from Utah with a bachelor's degree in sociology is sourced to an article written before he graduated and not mentioning his major: this for sure looks hallucinated. Struck out because shown to be incorrect, I'm sorry for the mistake. 02:54, 24 July 2026 (UTC)
I think your concern is entirely justified. I will warn the editor. M kuhner (talk) 03:08, 22 July 2026 (UTC)
Thank you, M kuhner for the comment on the dispute with WikiOriginal-9.
Per your questions:
- Yes, I did write this article. Have edited in the past but it has been many years...
- Local media (KSL.com) is considered a highly reputable news source. Maybe the original rivals rankings would be better as an authoritative original source, that's a fair criticism.
- The #4 and #1 rankings are both mentioned. They are just mentioned in different sections due to the chronological orders. He was the #4 recruit coming out of high school by Scout.com. Since it was a high school based honor, this was mentioned in the early life section. The #1 ranking was when he was coming out of Junior College and transferring to Utah, so it was mentioned in the college section.
- A sack is a loss of yards on the play, by definition. It's simply different phrasing.
- The graduating with a degree in Sociology is accurate. He graduated from high school in 2006 and left Utah in 2012--six years later. There are a few possible reasons for your confusion... athletes don't always leave school after graduation. They often stay in school and enroll in graduate school (no record of him being listed as a graduate student on the roster so not likely the case) or taking a second bachelor degree (more likely given his class is listed as senior and not graduate student). But there is also a possibility that the Utah Athletic Department (source cited) updated his bio once he did graduate. Either way, Utah Athletics is an authoritative source. ~2026-40064-42 (talk) 20:17, 22 July 2026 (UTC)
Sack and tackle-for-loss are not the same thing. ~WikiOriginal-9~ (talk) 20:24, 22 July 2026 (UTC)
The problem is that the article cites source A for information only present in source B. As I said above: "The source given for Aiono graduated from Utah with a bachelor's degree in sociology is sourced to an article written before he graduated and not mentioning his major". Even if the statement about his degrees is true, it's incorrect to give a citation to an article that does not support it. This is an extremely common error in LLM text, which is what caused my concern. Struck out as incorrect, see below. 02:55, 24 July 2026 (UTC)
That's the only serious error in this edit. My next step is to check if this is a pattern in other edits, which will take me a while. M kuhner (talk) 20:50, 22 July 2026 (UTC)
You said: "The source given for Aiono graduated from Utah with a bachelor's degree in sociology is sourced to an article written before he graduated and not mentioning his major".
That simply is not true. This is the source that is cited: https://utahutes.com/sports/football/roster/james-aiono/1262
It is his official Bio at the University of Utah, put out by the school. It explicitly says: "PERSONAL: Son of Ta'aloga and Ali'itasi Aiono ... youngest of five children ... married (Malvina) ... earned a bachelor's degree in sociology ... full name is James Pelenise Aiono (pronounced eye-oh-no) ... born Jan. 23, 1989 in Anaheim, Calif"
I am not sure what article you're talking about but the citation provided for the degree and major is an authoritative source that explicitly lists it.
Again, this was not written by LLM. It did not hallucinate any information. The edit should not be reverted.
If you have a problem with any part of the content, it should be improved. ~2026-40064-42 (talk) 15:11, 23 July 2026 (UTC)
You are right. I apologize wholeheartedly for my error and will strike it out above. I looked at the part that said he was a junior and assumed that therefore the article had been written before he graduated, but no excuse for not reading the lower part as well. M kuhner (talk) 02:53, 24 July 2026 (UTC)
No worries. I know mistakes happen. I appreciate your good faith efforts.
I put time into researching the articles and writing them over a few days. So it's a little frustrating to be accused of not putting in the effort and then having the works reverted.
Not sure why the WikiOriginal-9 wanted to blow away all my work because of the flow of the prose (which I did write and took a long time of reworking it to get the wording that seemed to work). ~2026-40064-42 (talk) 17:12, 24 July 2026 (UTC)

Haddad Maia fan

Latest comment: 30 days ago12 comments4 people in discussion
globalblock  Haddad Maia fan (talk · contribs · logs · block log) 
Editor has been globally blocked for AI/LLM policy violations. 

Haddad Maia fan (talk · contribs)

sizable case, drafts and comments exhibit signs like AILEGACY, SUPERFICIAL, RO3, misusing NSPORTS (check WP:NTENNIS vs their comment) (that last diff suggest that they've been using LLMs since 2024, ie most of their edits) ~2026-38678-64 (talk) 18:36, 21 July 2026 (UTC)

see also the ANI thread ~2026-38678-64 (talk) 18:37, 21 July 2026 (UTC)
Copying my comment from the ANI thread, the AI signs appear from the first creation, ~8000 edits and 101 creations btw:
CRESSPOM was created with 3 red-linked (ie. potentially hallucinated) templates, including a maintenance tag. Mostra tua força Brasil had/has plenty of WP:AIVOCAB like emphasized, shaped, highlighted etc and WP:SUPERFICIAL. Also see Luciana Mariano (sports journalist), Luiza Fullana, and Magali Fernandes. Kowal2701 (talk, contribs) 18:43, 21 July 2026 (UTC)
actually @Kowal2701, i think that it starts here (28 October 2024) because the other edits before then are quite small, don't seem to have signs, and are generally about adding 1 source or 1 image somewhere ~2026-38678-64 (talk) 18:50, 21 July 2026 (UTC)
Yeah my bad, I've only been looking at article creations , which started December 2024 Kowal2701 (talk, contribs) 19:42, 21 July 2026 (UTC)
So, how do thing proceeds from here. Tvx1 16:17, 27 July 2026 (UTC)
Their comment in the ANI thread didn't address what I said and was pretty evasive, also seemed AI, I guess we keep an eye on their future edits and check for LLM use. I'll respond to them and try to get a proper answer or commitment Kowal2701 (talk, contribs) 17:24, 27 July 2026 (UTC)
Can someone with more Portuguese knowledge compare "Mariano was married to sports broadcaster Luciano do Valle (1947–2014), who played a significant role in encouraging and supporting her early career in sports narration" to the source? My machine translation suggests the source says the opposite to the claim. CMD (talk) 09:57, 28 July 2026 (UTC)
Yeah, the source actually says the opposite. It's mostly an interview with her during which she states that her former husband did not push her into her career, but that it was purely her own choice and here own achievement.Tvx1 13:18, 28 July 2026 (UTC)
The very title seems to say the opposite. That seems a sign sources aren't being checked. Magali Fernandes still has markdown in the article text. CMD (talk) 00:39, 29 July 2026 (UTC)
So, the user was globally blocked as the ANI investigation found out that they were a sockpuppet of another glocked account. That means they can’t add more AI stuff to our articles, but we still need to look at cleaning up what they already added to Wikipedia.Tvx1 18:55, 31 July 2026 (UTC)
I'm tagging their articles for WP:G5 now Kowal2701 (talk, contribs) 13:12, 2 August 2026 (UTC)

Talk:Battle of Lumë#RFC on Infobox for Battle of Lumë

Latest comment: 20 days ago3 comments2 people in discussion
NO ACTION TAKEN
This problem has metastasized onto other noticeboards and is probably best left to them. (non-admin closure) M kuhner (talk) 19:39, 12 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Red X symbolN Cleanup is unnecessary.

Could someone take a look at the RfC above please? I've !voted so I can't, but there seems to be some heavy LLM usage going on. InfernoHues (talk) 19:15, 27 July 2026 (UTC)

That is an unpleasant RfC for sure. An uninvolved editor with good LLM detection skills could collapse the LLM text and see if that allows the discussion to progress--occasionally it does. I'm not quite confident in my skills here but one of the more experienced people could probably do it. If that does not solve the problem you'll probably need to involve ANI or AE. M kuhner (talk) 19:56, 27 July 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Autism articles

Latest comment: 25 days ago7 comments5 people in discussion

A new account is claiming at Autism in China that "Large chunks of this article are likely bot-generated". Could someone look over the Autism in China article? On the one hand, this editor apparently believes that I'm a bot, so I'm can't give their comment much credence, but on the other hand, it's exactly the kind of article (e.g., about a non-English-speaking country) where we could expect to see editors resorting to LLMs for copyediting. WhatamIdoing (talk) 03:37, 30 July 2026 (UTC)

Only possible substantial AI edits would be these (everything else after 2022 is minor); not really getting any indicators of AI from them. If there are reference problems that's still an issue of course. Gnomingstuff (talk) 03:58, 30 July 2026 (UTC)
I spotchecked the citations. Source author Yang Wiwei's name appears in the citation as "Tone, Sixth" which may be a machine translation failure. The Zhou citation leads to an article whose credibility I doubt (but this wouldn't necessarily be clear to a lay person or non-native speaker). But every one I checked goes to a real and apparently relevant source, which would be unusually good for LLM, and I did not find source/text mismatches. (I checked about 10 sources, avoiding Chinese and paywalled ones. Usually that's enough to see problems.)
The sourcing needs a lot of work, but there's a decent article possible here if someone can clean it up.
To elaborate on Gnomingstuff's point, it's not possible that large chunks are bot-generated, because most of the text is too old; the article was written in 2021. M kuhner (talk) 04:06, 30 July 2026 (UTC)
Yeah thanks for the clarification, what I meant was those are the only edits that made anything beyond minor changes to the text, and they don't seem like AI. Although if someone says "bot-generated" that kind of sounds like something that isn't an LLM, like some kind of mass-written thing? Gnomingstuff (talk) 05:35, 30 July 2026 (UTC)
Or it could just mean that anything an editor disagrees with is "obviously" [fill in the blank with your choice of procedural reasons to remove it or ignore other editors]. WhatamIdoing (talk) 17:38, 30 July 2026 (UTC)
Noting that Sixth Tone (RSP entry) is the name of a magazine. — Newslinger talk 22:16, 7 August 2026 (UTC)
Without seeing the other editors' comments, I came to the view that the series of edits made in April 2024 are problematic, particularly in the "Societal stigma" section. Fuzzy, generalised wording with poor sourcing that's only vaguely connected to the draft's text. ChrysGalley (talk) 08:09, 30 July 2026 (UTC)

User:Howeteher23

Latest comment: 29 days ago2 comments2 people in discussion

User has made a bunch of 10k+ edits in the span of an hour earlier today. I already reverted their edit on Standing on the Shoulder of Giants as it's on my watchlist and I firmly believe these are AI edits. – zmbro (talk) (cont) 18:18, 2 August 2026 (UTC)

I've undone the rest of their edits. msk 00:28, 3 August 2026 (UTC)

User:Sdylans

Latest comment: 18 days ago6 comments3 people in discussion

New user, but has created two articles and made large edits to others. Should be nipped in the bud before it gets worse. Mostly edits BLPs for Slovenian artists.

Has also made large edits and expansions to other articles:

Hopefully we can win ourselves a new working Wikipedian. Many of our articles on Slovenian artists seem to be in a sorry state and can definitely use improvements. ‑‑gurkubondinn 12:47, 5 August 2026 (UTC)

The hallucinated source in Vid Simoniti can be found at https://www.liverpool.ac.uk/people/vid-simoniti . The first paragraph of the Academic career section appears to be closely paraphrased from this source. ClumsyOwlet (talk) 06:40, 6 August 2026 (UTC)
Sure, I also tried replacing the m. with www., but thats not the point. The URL that was cited does not exist, because it was hallucinated by an LLM that incorrectly predicted/generated the URL. If the user had actually been reading that source themselves and had written their contribution themselves, they would have copied the URL from their browser where they were reading it. But they weren't actually reading the source, so they didn't copy the URL and instead we got this incorrectly generated URL by the LLM that generated the contribution that the user pasted in here. The incorrect URL is a symptom, it is not the problem itself. ‑‑gurkubondinn 13:25, 12 August 2026 (UTC)
...did you mean to send this reply in #User:Weichan123? ClumsyOwlet (talk) 04:51, 14 August 2026 (UTC)
No, it's a reply to you. I mis-remembered the URL structure though. Not sure where I was lookig at an m. domain, but I guess I got those two mixed up. But I was aware that the page existed at a different URL, but that wasn't the point. ‑‑gurkubondinn 10:35, 14 August 2026 (UTC)
vanished as Renamed user 311109cfefb0b593d7ae477ff005460b (talk · contribs), I've done the remaining edits Kowal2701 (talk, contribs) 10:32, 14 August 2026 (UTC)

Proposed new user talk page template

Latest comment: 25 days ago2 comments2 people in discussion

 Template Link: Template:Uw-aiask

 Main Discussion Link: Wikipedia talk:Template index/User talk namespace § New User Warning Template

I have made a new talk page template related to AI use and am attempting to get it implemented. I thought that it would be beneficial to have an user talk page template to deal with potential AI misuse that doesn't presume AI usage, but instead, asks if they've used it. I believe it has many use cases in situations where one suspects that AI misuse but is still unsure. Even in situations where you are postive that an user has misused AI, it can be beneficial (especially towards newer users) to make a less confrontational approach. I would appreciate feedback from people that regularly interact with LLM misuse. Feel free to leave your thoughts here or comment on the main discussion. Thanks! - Otherwise (Antibacklog Aktion!) (talk?) 15:23, 5 August 2026 (UTC)

I used to have a de facto template asking whether someone was using AI, what their prompts/workflow/etc was, and what if any review they did of the output, but I stopped using it since in my experience people just ignored it, and even if they didn’t very few people actually shared that information in full beyond vague weasel wording Gnomingstuff (talk) 21:23, 7 August 2026 (UTC)

Luiza Prado

Latest comment: 18 days ago4 comments3 people in discussion

Hi, I am concerned that this big edit here at the Luiza Prado article which is the first edit of a new user is LLM generated. I'm just going by the stilted writing, grammatical perfection, and constant peacock wording, imv Atlantic306 (talk) 00:59, 9 August 2026 (UTC)

Yup, looks like an AI duck to me. Compare "Prado's work ... centres on memory, perception, neurodivergence..." and WP:AILEGACY, or even the sentence "Photography has a structural role in Prado's practice...". I'm not sure what to make of this page, which shares only its name with the above article. I'm a bit worried that this person is a promotional editor, and is LLM-generating this page with the sole intention of having a hatnote redirect to Luiza Prado. Either that, or they're simply creating random articles on a whim, and their LLM found the second subject for them. Somepinkdude (talk | contribs), in solidarity 03:03, 9 August 2026 (UTC)
This  megaphone Sounds like an AI chatbot beeping into a megaphone to me Somepinkdude (talk | contribs), in solidarity 16:24, 9 August 2026 (UTC)
edits reverted and drafts tagged, worth keeping an eye on Kowal2701 (talk, contribs) 11:42, 14 August 2026 (UTC)

User:ThefirstmanofThai

Latest comment: 22 days ago3 comments2 people in discussion
siteblock  ThefirstmanofThai (talk · contribs · logs · block log) 
Editor has been blocked site-wide for AI/LLM policy violations. 

The nature of their edits and the large changes in very short periods of time show multiple WP:AISIGNS. This is NOT my area of expertise, but this warrants an in-depth look by someone with more experience with AI. Zackmann (Talk to me/What I been doing) 21:39, 9 August 2026 (UTC)

Courtesy ping: Aesurias Zackmann (Talk to me/What I been doing) 21:44, 9 August 2026 (UTC)
Edit summaries are as clearly AI-written as can be, and sometimes fail to match the actual edit (e.g. Diff/1368158423 claims to expand "historical context" but mostly just deletes information). They also add tons of naked citation templates without the <ref> tag (and to cite a source again they repeatedly reproduce the entire template), which might fall under WP:AICITESTYLE and in any case has made a huge mess on a number of articles. And they appear to be ignoring the attempts to reach out to them on their talk page. Astro.furball (talk) 00:54, 10 August 2026 (UTC)

Te Paea Paringatai

Latest comment: 22 days ago5 comments4 people in discussion

I've come across this article and the wording is a bit strange. Especially this line "This should take into account the knowledge and the culture of the Maori people which is known as Mātauranga Māori. Her upbringing and heritage are helpful here.". I'm not an expert but this sounds very AI to me, could someone more versed in LLMs take a look at it please? The C of E God Save the King! (talk) 09:41, 10 August 2026 (UTC)

tbh it doesn't read as AI at all to me, and this is Victuallers so extremely unlikely to be. It is editorialising though Kowal2701 (talk, contribs) 09:56, 10 August 2026 (UTC)
Not LLM, I can rephrase that sentence to be more encyclopedic, however. Sarsenet•he/they•(talk) 10:23, 10 August 2026 (UTC)
Thank you. AI is better at phrasing than me! Victuallers (talk) 10:37, 10 August 2026 (UTC)
My apologies @Victuallers:. I didn't know you were the author and when I saw that phrase, it just didn't look right to me so that's why I thought it was AI from my limited experience of LLMs on noticeboard discussions. The C of E God Save the King! (talk) 10:53, 10 August 2026 (UTC)

User:Julien Sow 2

Latest comment: 1 month ago1 comment1 person in discussion

The user has a habit of pasting AI-generated articles into their sandbox, and then moving the sandbox out into article space. Found them pasting chatbot output in Diff/1364169939 (after the page is deleted, the trailing marker can still be seen in AbuseLog/44652125), and looked through three of their pages and found AISIGNS in all of them:

There is extremely rapid editing, with gigantic edits that can't possibly have been read properly in the timeframe (let alone written).

The user has promised to stop using LLMs in § Speedy deletion nomination of User:Julien Sow 2/sandbox on their talk page. I don't know if this can be cleaned up or should be deleted altogether. More eyes on this is appreciated. ‑‑gurkubondinn 10:55, 15 July 2026 (UTC)

User:Jsinnyj

Latest comment: 17 days ago16 comments8 people in discussion

Jsinnyj (talk · contribs · deleted contribs · logs · filter log · block user · block log)

Continually removing AI tags in rather hostile fashion, claiming they are not using AI. Edits like Special:Diff/1223722511 and Special:Diff/1222630998, I feel, speak for themselves. (Also look at timestamps; for instance, the latter was added 3 minutes after the previous edit.)

I do not think that all of the text is AI-generated; for instance, the last paragraph of Special:Diff/1235942581 not only doesn't sound like AI but has some URL parameters consistent with various Firefox search actions (it looks like this bug) that I'm not familiar with chatbots spoofing. It seems like the AI edits, for some reason, are interspersed in. Gnomingstuff (talk) 22:08, 17 June 2026 (UTC)

And what is going on at User talk:Sherlockian1234 § Managing a conflict of interest? Is Jsinnyj responding to a COI notice for someone else, or am I just missing something? ‑‑gurkubondinn 11:33, 18 June 2026 (UTC)
Hi jsinnyj here...can you explain to me what the issue is? How am I hostile? Jsinnyj (talk) 14:28, 8 July 2026 (UTC)
Why did you reply to a COI template that was placed on a different user's talk page? Do you have any connection to the Sherlockian1234 user? ‑‑gurkubondinn 14:45, 8 July 2026 (UTC)
And in 2023, the user was removing COI templates from the same page: Diff/1140786842. This is also pretty much the only article that this user has ever shown interest in editing. ‑‑gurkubondinn 20:48, 4 August 2026 (UTC)
the user is still doing it, a month later? ~2026-38678-64 (talk) 23:07, 19 July 2026 (UTC)
They haven't edited since the 8th July and seem to be developing a pattern of taking long periods away from Wikipedia and then coming back just to remove the AI-generated tag and disappear again. I have no idea how we're supposed to pin this one down, since admins don't block inactive users. I wonder if they would consider p-blocking from just the Jason Barr page? purringmaggot (talk) 23:44, 19 July 2026 (UTC)
Rinse and repeat until someone catches them not too long after they come back and manages to convince an admin to pull the plug? ‑‑gurkubondinn 23:53, 19 July 2026 (UTC)

The user page no longer exists. But for future for any editor or other instances it would seem reasonable to leave warnings on their talk page and delete their content altogether. This seems like it would help build a stronger and more urgent case for WP:DE. AadamentAardvark (talk) 18:39, 4 August 2026 (UTC)

They have not edited since July 8: I propose a community consensus for cleanup. We will want to do it carefully as they seem to have used LLMs only sporadically. I have made a tracker. M kuhner (talk) 00:35, 15 August 2026 (UTC)

Support. ‑‑gurkubondinn 00:42, 15 August 2026 (UTC)
I looked in detail at their additions to New Mannahatta, which have been transferred to a section in Land_reclamation_in_Lower_Manhattan. All sources exist and say what they are claimed to say. There is a little LLM-style vagueness but not really enough to nail it for me.
However there is a pronounced non-neutrality toward Jason Barr (the subject of this editor's attention). Opposition to Barr's plans is dismissed in half a sentence, followed by a paragraph on why Barr is right. There is some SYNTH or OR involved in the use of general references to support Barr's views. This plus the weird interaction with Sherlockian1234 mentioned above leads me to suspect COI. Probably what this section needs is to be cut down to facts.
The Jason Barr article is also quite pro-Barr. We are told both that 'skynomics', skyscraper economics, is a field, and that hardly anyone but Barr is in it. It also has at least 2 broken sources (I stopped checking after that) and the LLM tic of putting a reference mid-paragraph and then adding several lines or a whole paragraph of unsourced material after it. One unsourced block suddenly switches from "Barr" to "Dr. Barr" and has probably been taken from elsewhere. Barr is likely notable but this article is questionable. I wonder if stubification is the best answer here. M kuhner (talk) 01:29, 15 August 2026 (UTC)
The "Dr. Barr" section is 4 lines of extremely close paraphrase from the subject's CV. It's pretty rote stuff though. M kuhner (talk) 01:36, 15 August 2026 (UTC)
didn't see this sorry, I just nuked it Kowal2701 (talk, contribs) 21:07, 15 August 2026 (UTC)
Deleting my previous comment as I assumed "nuked" was a bit more destructive than the actuality. Good cut-downs in both cases, thank you. The cleanup is done but the user is not blocked. Close now, or keep it around to remind us to check whether they've come back? M kuhner (talk) 23:06, 15 August 2026 (UTC)
I'll catch them in the archives :) Kowal2701 (talk, contribs) 23:10, 15 August 2026 (UTC)

Lumberjack essay

Latest comment: 18 days ago4 comments3 people in discussion
MOVED
The search continues at WT:AIC. (non-admin closure) M kuhner (talk) 20:22, 14 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

There was a good essay somewhere about the pitfalls of LLM-assisted proofreading somewhere. It featured an example about rewriting a short passage about an unnamed lumberjack. –LaundryPizza03 (dc̄) 00:18, 12 August 2026 (UTC)

@LaundryPizza03: you might want to move this to WT:AIC instead? ‑‑gurkubondinn 15:56, 12 August 2026 (UTC)
Wikipedia_talk:WikiProject_AI_Cleanup#Lumberjack_essay_about_AI_proofreading. –LaundryPizza03 (dc̄) 16:08, 12 August 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

User:PublicRecordEditor

Latest comment: 17 days ago27 comments8 people in discussion
BLOCKED
blocked as sockpuppet per this investigation. Pages have been tagged for G5 speedy deletion. - Otherwise (Antibacklog Aktion!) (talk?) 21:27, 15 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Green checkmarkY Cleanup has been completed.

PublicRecordEditor (talk · contribs · deleted contribs · logs · filter log · block user · block log)

New user with prior history of llm usage (```wikitext) that is rapidly creating new mainspace articles (4 new articles in mainspace in a 45 minute time span, all more than 2800 bytes ). Would appreciate help gauging if their new articles are written with AI. they are relatively short and don't have any super obvious indicators that I could see, but I might've missed something. - Otherwise (Antibacklog Aktion!) (talk?) 19:57, 12 August 2026 (UTC)

AI edit summaries plus this whole article written in about 1 minute = WP:DUCK.
(The most obvious indicator here is the edit summaries, there's a very narrow summary template style that is like the tic of AI these days) Gnomingstuff (talk) 20:03, 12 August 2026 (UTC)
Do you think it warrants G15? I dont know how "without human review" applies in practice - Otherwise (Antibacklog Aktion!) (talk?) 20:08, 12 August 2026 (UTC)
I wrote these articles without the help of AI. I wrote them in Google Docs and used grammar and spelling tools to clean up the text before publishing. I understand the concern, but would be glad to answer any queries you may have regarding my editing process, or my sources. PublicRecordEditor (talk) 20:17, 12 August 2026 (UTC)
used grammar and spelling tools
What tools? Gnomingstuff (talk) 20:23, 12 August 2026 (UTC)
Google Docs' built-in spelling and grammar checker. I used it only for spelling and grammar corrections not to generate or rewrite t he article text. PublicRecordEditor (talk) 20:26, 12 August 2026 (UTC)
Did you ever write your edits in markdown? Or are you unfamilar with markdown. - Otherwise (Antibacklog Aktion!) (talk?) 20:34, 12 August 2026 (UTC)
I know Markdown, but I don't usually write Wikipedia edits in Markdown. For small edits I usually edit directly in Wikipedia's Source Editor, using wikitext. For longer drafts I write the prose in Google Docs first and copy it over to Wikipedia, then use the editor and preview to format and check it before publishing. PublicRecordEditor (talk) 20:43, 12 August 2026 (UTC)
Please don't lie about your AI usage. This edit of yours has ```wikitext at the top and ``` at the bottom which is a known sign of LLM generation. I understand that as a new editor, it can be frustrating to have your changes reverted and your editing scrutinized, but not being truthful about your AI usage is likely to exacerbate the situation. You may find this essay useful. Please let me know if you've got any questions. - Otherwise (Antibacklog Aktion!) (talk?) 20:54, 12 August 2026 (UTC)
Thanks for explaining and sharing those links. I understand your worry and appreciate your aid. Thanks again. PublicRecordEditor (talk) 21:16, 12 August 2026 (UTC)
Otherwise I share your concern regarding PublicRecordEditor's use of AI chatbots and I agree that editors should be upfront about how their contributions are prepared. The concerns outlined at WP:LLM including inaccurate information and fabricated references more unreviewed LLM generated text should be taken seriously. However PublicRecordEditor appears to be a new user who may not yet understand Wikipedia's formatting/sourcing and AI LLM related guidance.WP:AGF asks editors to assume good faith unless there is clear evidence to the contrary.while WP:BITE encourages patience and constructive guidance for newcomers We can correct or revert problematic contributions while also helping the editor understand what went wrong. Peakspire (talk) 21:59, 12 August 2026 (UTC)
eye I have read the above message. I will reply when I have a moment. - Otherwise (Antibacklog Aktion!) (talk?) 00:33, 13 August 2026 (UTC)
Dear @Peakspire,
Thank you for your comment. I agree with you that WP:AGF and WP:BITE are incredibly important and that we need to be accommodating to newcomers. In fact, WP:BITE is one of the standards that I'm most passionate about on Wikipedia. However, in this case, I don't feel that my actions have violated WP:BITE.
I don't think that it's contested to say that the editor in question has used an LLM. Their history shows clear LLM usage, and their recent actions show several signs of LLM generation. This in and of itself is not a condemnation of the editor. A brief look at the archives of this noticeboard show that instances of AI usage by an editor are usually resolved not with a sanction, but with a conversation with the editor, an acknowledgement of LLM misuse, and a commitment to not do so in the future. I think that I've made it clear that LLM usage is not allowed, but thus far, they have not even acknowledged that they've used it.
WP:BITE makes sure that newcomers don't have to know all the rules before they make an edit, but newcomers must be able to acknowledge their mistakes and attempt to avoid them in the future. To quote from an essay that I wrote:

WP:BITE is NOT a shield from criticism. If any editor, including a newcomer, makes a mistake, there should be no hesitation to point out the mistake...It is one of the best ways that a newcomer can learn.... Attempts to use bite solely to refute criticism are harmful.

I'd like to take the opportunity to invite @PublicRecordEditor to commit to not misuse LLMs in the future. I hope this incident doesn't drive you away from making constructive edits in the future and if there's anything that I can help clarify, feel free to let me know. - Otherwise (Antibacklog Aktion!) (talk?) 01:30, 13 August 2026 (UTC)
I insist that the articles recently discussed here are written by me and not generated by an LLM. I do acknowledge that I used LLM tools before and I understand the concerns that came from it. I have read the guidance and will not use LLMs in ways that violate Wikipedia's rules in the future. PublicRecordEditor (talk) 02:08, 13 August 2026 (UTC)
That's a pretty vague commitment. You are prohibited from using an AI as an author or co-oauthor of any text on Wikipedia, whether it's an article or talk page comment. You can use it for finding sources or offering advice, and maybe translation, but that's it. If you are incapable of functioning on Wikipedia without AI help, then you need to find some other activity with which to spend your time. ~Anachronist (who / me) (talk) 15:16, 13 August 2026 (UTC)
Thanks for that, Anachronist. Let me be perfectly clear about the earlier draft that was referenced. I did use Quillbot when working on that older draft. This was my first attempt at creating an article while I was still learning my way around Wikipedia. I understand, being new and the article still being in draft space does not forgive that mistake. Sorry for the use of the tool and the confusion it caused. When I realized the issue, I ceased the use of Quillbot or other LLMs to write or rephrase Wikipedia content. I'll start writing my own article text, talk page comments, and noticeboard comments from now on. I will not list an LLM as an author or co-author of my contributions to Wikipedia. I understand the rule now, and I'll do it. If I'm unsure about something in the future, I'll ask other editors for guidance before making the edit. Sorry for the earlier mistake and the extra work it caused. Thanks for the opportunity to clarify this. PublicRecordEditor (talk) 16:47, 13 August 2026 (UTC)
In case there's any doubt (cc: @Peakspire, per your comment on the case page), they've definitely used Quillbot, an AI humanizer, before. The smoking gun is Special:diff/1368252303. In it, PublicRecordEditor seems to have left in the lines
"Rephrase sentence
Revert to original sentence
Copy sentence
Synonyms
House
[etc..]"
If you go to Quillbot's AI humanizer, paste in a random sentence, click humanize, and click on words in the humanized sentence, you will find the exact same lines, just with different synonyms. ARandomName123 (talk)Ping me! 05:53, 13 August 2026 (UTC)
damn I didn't know that "Google Docs' built-in spelling and grammar checker had that
I don't understand why people don't just answer a simple question with the truth Gnomingstuff (talk) 14:43, 13 August 2026 (UTC)
Thanks for this find ARandomName123
Using an AI humanizier after your Articles for Creation submission was denied specifically for LLM use implies that you knew that AI use was prohibited, and nevertheless, didn't disclose it and instead took action to obscure it.
@PublicRecordEditor, did you really not use AI in your most recent articles, and if so, why did you suddenly stop using AI? It's difficult to reconcile your actions with your words when you are not substansively engaging about why and to what extent you used AI. I understand that it can be difficult for a newcomer to approach Wikipedia, but your actions thus far haven't exactly been forthcoming. I would highly encourage you to be honest as the community's patience, even towards newcomers, has its limits. - Otherwise (Antibacklog Aktion!) (talk?) 16:11, 13 August 2026 (UTC)
Hi, Otherwise. Following the AfC decline, my previous explanation is difficult to accept and I can see why the Quillbot edit makes it seem that way. I want to answer your questions honestly. I did use Quillbot on that earlier draft, when I was still learning how to write articles on Wikipedia. In retrospect, I do think it was a bad idea to use it given the AfC concern, and I should have been more open about it when you first asked me about my tool usage. Sorry about the last one. I stopped when I understood that using AI to write or rewrite Wikipedia content was not okay. Since then I have been writing the recent articles from the sources I research myself. I didn’t use Quillbot, ChatGPT, or another LLM to generate or rewrite the text for the recent articles. I do not try to cover up my previous mistake or evade responsibility for it. The use of the Quillbot being discussed was on the earlier draft and I can see why that raised concerns. I will not be using an LLM in the future as author or co-author of article text, talk page comments, or noticeboard comments. I am still learning Wikipedia processes, but I understand this rule now and will abide by it. Sorry I wasn't more clear about the earlier use in my first response. PublicRecordEditor (talk) 16:49, 13 August 2026 (UTC)
Thanks for the answers. Could you explain how the Diff/1368810661 edit was produced and what happened there? ‑‑gurkubondinn 23:15, 13 August 2026 (UTC)
Thanks for asking. To answer directly: Yes, that edit (Diff/1368810661 to Draft:Ans Ishfaq) was drafted with help from ChatGPT. I worked on the text in Google Docs before posting it, and I missed removing the ```wikitext``` formatting markers when I pasted it in. That's on me. This falls in the same window I've already described here: using AI tools before I fully understood the norms around AI-assisted writing on Wikipedia. I've stopped that. Going forward, I won't use AI to draft or rewrite article text or discussion comments on this draft or anywhere else. I'm glad to go through the affected sections of the draft now and rewrite them myself, in my own words, from the sources already cited. PublicRecordEditor (talk) 02:20, 14 August 2026 (UTC)
Is it just me or does this reply look AI generated? - Otherwise (Antibacklog Aktion!) (talk?) 18:00, 14 August 2026 (UTC)
Partly, maybe. I have to wonder why the word "wikitext" was surrounded by markdown code delimiters. ~Anachronist (who / me) (talk) 20:37, 14 August 2026 (UTC)
Maybe I'm misunderanding you, but Im assuming that's because they're responding to a comment by gurkubondinn asking about why the markdown delimiters and the wiki text tag were present in this diff. - Otherwise (Antibacklog Aktion!) (talk?) 21:10, 14 August 2026 (UTC)
Unsurprisingly, this was a sock . It looks like Izno has G5'd and reverted everything from PRE and another AI using sock Bhatti76199, so this one is done NicheSports (talk) 20:15, 15 August 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

User:ElijahJWrites

Latest comment: 1 month ago5 comments3 people in discussion

I think I need another eye on this user to determine if they have a history of AI edits to clean up. I spotted the LLM use at this diff which added this amusingly vague, almost information-free section: . However, many of the user's other contributions, despite having obviously AI-generated edit summaries, introduce typos and ungrammatical punctuation usage (see where the edit summary ironically states that it corrected spelling errors and artist name inconsistencies while introducing no less than 5 misspellings of the subject's name). I don't know what to make of the distinction between the AI summaries and the, let's say, organically spelled content, and the potential mixing of human and LLM content within a single edit (see which edit summary boasts neutral, concise, and source-based Wikipedia-style wording and adds a recognizably AI-generated Awards & Recognition section but also adds a Television appearances section bizarrely written in the first person).

Anyway, I warned the user for the LLM usage in the first diff and tagged the article (looks like the whole thing was already AI-generated before they got there) purringmaggot (talk) 02:08, 17 July 2026 (UTC)

I don't know what to make of stuff like led to several more appearances that involved a close interaction with the host, that is sourced to an episode of Super Dave on Apple TV..? ‑‑gurkubondinn 12:32, 17 July 2026 (UTC)
* 1990:The Arsenio Hall Show is already mentioned in the entry, but I did not perform routines. I performed only one routine: my signature Ballon-man act. sounds like some "ask Character AI if the article is accurate" workflow, maybe? I don't know how you'd even get this otherwise, outside some kind of weird larping or "Elijah J" actually being Blumenfeld but not a SPA Gnomingstuff (talk) 13:48, 17 July 2026 (UTC)
User has now denied LLM use. Personally I find this unconvincing and I've said so; we'll see how this develops. purringmaggot (talk) 15:41, 17 July 2026 (UTC)
Now stating LLM was only used for "formatting" and a not-quite-promise to "be very careful with this next time": (the conversation got split up between my talk and theirs) I think it might also be worthwhile to keep an eye on this user Kappalotti who originally made the AI-generated Dwight Cuthbert Revire Gardiner page and has come out of inactivity as of yesterday. Whatever's going on here, I'm at least definitely going to stubify that page whenever I have the time. I'm supposed to be at work! purringmaggot (talk) 16:18, 17 July 2026 (UTC)

User:Aerogabs

Latest comment: 15 days ago4 comments3 people in discussion
CLEANUP COMPLETE
Editor appears inactive. (non-admin closure) M kuhner (talk) 01:14, 17 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Green checkmarkY Cleanup has been completed, see /2026-08-07 Aerogabs for reference.
notified  Aerogabs (talk · contribs · logs · block log) (7 August 2026)
Editor has been notified and asked to discuss AI/LLM editing concerns. 

Editor making several seemingly promotional edits to wrist watch related articles. Some of their recent edits are pretty blatantly promotional and the edit summaries are screaming AI. I'm unsure to what extent they used AI and they haven't responded to any talk page warnings thus far. Would appreciate help investigating their edits and potentially reverting if needed. - Otherwise (Antibacklog Aktion!) (talk?) 14:30, 7 August 2026 (UTC)

That's not the first LLM user that I've come across on watch related articles, there was a pretty persistent one on the Breitling article some time ago. ‑‑gurkubondinn 12:18, 13 August 2026 (UTC)

They have not edited since August 7. Hopefully that's the end of it. Closing as completed, but noting there has been no block. M kuhner (talk) 01:14, 17 August 2026 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

User:ElijahJWrites

Latest comment: 1 month ago5 comments3 people in discussion

I think I need another eye on this user to determine if they have a history of AI edits to clean up. I spotted the LLM use at this diff which added this amusingly vague, almost information-free section: . However, many of the user's other contributions, despite having obviously AI-generated edit summaries, introduce typos and ungrammatical punctuation usage (see where the edit summary ironically states that it corrected spelling errors and artist name inconsistencies while introducing no less than 5 misspellings of the subject's name). I don't know what to make of the distinction between the AI summaries and the, let's say, organically spelled content, and the potential mixing of human and LLM content within a single edit (see which edit summary boasts neutral, concise, and source-based Wikipedia-style wording and adds a recognizably AI-generated Awards & Recognition section but also adds a Television appearances section bizarrely written in the first person).

Anyway, I warned the user for the LLM usage in the first diff and tagged the article (looks like the whole thing was already AI-generated before they got there) purringmaggot (talk) 02:08, 17 July 2026 (UTC)

I don't know what to make of stuff like led to several more appearances that involved a close interaction with the host, that is sourced to an episode of Super Dave on Apple TV..? ‑‑gurkubondinn 12:32, 17 July 2026 (UTC)
* 1990:The Arsenio Hall Show is already mentioned in the entry, but I did not perform routines. I performed only one routine: my signature Ballon-man act. sounds like some "ask Character AI if the article is accurate" workflow, maybe? I don't know how you'd even get this otherwise, outside some kind of weird larping or "Elijah J" actually being Blumenfeld but not a SPA Gnomingstuff (talk) 13:48, 17 July 2026 (UTC)
User has now denied LLM use. Personally I find this unconvincing and I've said so; we'll see how this develops. purringmaggot (talk) 15:41, 17 July 2026 (UTC)
Now stating LLM was only used for "formatting" and a not-quite-promise to "be very careful with this next time": (the conversation got split up between my talk and theirs) I think it might also be worthwhile to keep an eye on this user Kappalotti who originally made the AI-generated Dwight Cuthbert Revire Gardiner page and has come out of inactivity as of yesterday. Whatever's going on here, I'm at least definitely going to stubify that page whenever I have the time. I'm supposed to be at work! purringmaggot (talk) 16:18, 17 July 2026 (UTC)

User:Wickingwiki

Latest comment: 18 days ago3 comments2 people in discussion

Wickingwiki (talk · contribs · deleted contribs · logs · filter log · block user · block log)

Added a plot summary to my article, Jimmy and the Pulsating Mass, which upon closer look has several AISIGNS hits, including WP:AIPUFFERY: the vibrant, lighthearted facade of the world, WP:SUPERFICIAL: gaining distinct abilities to overcome obstacles and fight off the infestation, rapidly destroying his brain and erasing his memories, WP:RO3, some vocab (utilizes, vibrant), as well as what I perceive to be factual errors (from my reading of non-rs on the plot, as I have not finished the game, it appears to be extrapolating quite a bit).

What sealed this for me was this, however: diff: Here is the section cleanly formatted in standard Wikipedia wikitext syntax:

I wouldn't normally open a case for something as obvious as that last diff, but I'm slightly involved as my article was affected. ASUKITE 02:40, 14 August 2026 (UTC)

The edit summaries are typical LLM throughout, as are quirks of the table formatting (a human who knows about non-breaking spaces generally also knows they aren't needed on very short texts). I spot-checked two sources added to Saints Cheer Krewe and neither reached its intended article. In other words, you may be involved but you are certainly not wrong.
I am also perplexed by an apparently new editor who removed 200,000 bytes from an article on their twelfth edit. I expect we've seen this person before. I was pretty bold as a newcomer but not that bold. M kuhner (talk) 02:59, 14 August 2026 (UTC)
They admitted to it on their talk page, would have saved some trouble if I'd waited. I'll just await their reply to decide what to do next. Thanks! ASUKITE 14:31, 14 August 2026 (UTC)

Request for second opinion on collapse of talk-page comments as LLM-generated

Latest comment: 16 days ago11 comments8 people in discussion
RESOLVED
Editor was indefed after continued LLM-enabled soapboxing elsewhere (non-admin closure) purringmaggot (talk) 17:15, 16 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

I would like to ask a second opinion about 2 discussions I opened at Talk:Sex differences in intelligence

Both discussions were collapsed as LLM-generated. However, the substantive arguments, the studies I selected, and the points I wanted to raise were my own. English is not my native language, and I used ChatGPT mainly to help improve the wording and clarity of what I had already decided to say.

My understanding of WP is that there is a distinction between comments that are generated by an LLM and comments where an LLM is only used to refine the expression of an editor's own ideas.

The two discussions are:

Talk:Sex differences in intelligence

Adult WAIS and Raven evidence warrants less categorical wording

WAIS and Raven's male advantage

I have not removed the collapse templates myself because I would prefer to ask for uninvolved input first.

Could editors here give their opinion on whether collapsing these discussions as LLM-generated is appropriate under the current guidance?

Thanks. ~~~~

Mill-John (talk) 08:49, 14 August 2026 (UTC)

The relevant guideline is WP:AITALK and there is an elaboration at the information page WP:LLMCHAT. Technically speaking if you used LLM to refine expression of your own ideas, you fall within the permitted uses under this guideline.
However, many Wikipedia editors have bitter experience with the tendency of LLMs to misrepresent sources. So when you post a lengthy list of sources accompanied by LLM text, people can't take them at face value. They don't know you're an expert: everyone sounds like an expert once LLM gets involved. They don't know if your summaries of the articles are real or LLM hallucinations. They also don't know if you are going to actually read and consider their responses, or respond with automated text. It's hugely damaging to your perceived credibility.
You are technically in the right and could unhat those comments. But you won't be able to get them taken seriously. I strongly recommend using your own words. They may be less polished, but that's okay. M kuhner (talk) 12:48, 14 August 2026 (UTC)
The examples given for what counts as refining expression are "check their grammar" and "translate words they are unfamiliar with". The collapsed sections seem to go far beyond that. CMD (talk) 12:59, 14 August 2026 (UTC)
Unfortunately those are the examples in WP:NOLLM, the mainspace article guideline. There's nothing like that in the comments guideline: it only says that comments obviously generated (not merely refined) by a large language model can be collapsed, and does not define "refined" at all. (I would support a proposal to change that. LLMs are toxic in comment threads; we've had several such cases recently. And while I can tell LLM from not-LLM fairly well, I can't tell generated from refined with any reliability.) M kuhner (talk) 17:42, 14 August 2026 (UTC)
Wikipedia:Large language models#Communicating applies to talkpage comments, as currently written and framed. CMD (talk) 23:43, 14 August 2026 (UTC)
I'd second what M kuhner said about credibility - our tolerance for poor grammar or non-native style is very high, but our tolerance for AI-processed text is very low. You will always be taken more seriously if you write in your own voice. As a last resort, you could try posting both: post your thoughts in your own words, and add the AI-processed version in a collapse box in case anyone finds it helpful to clarify your meaning. -- LWG talk (VOPOV) 23:56, 14 August 2026 (UTC)
  • A bit of context here: the OP created their account the same day that this article needed to be page protected (by ScottishFinnishRadish) because a temp account was edit warring to include quite similar content. Indeed, the only thing differentiating the edits of that temp account from this named account was the named account's apparent fluency in English. The insistence that we qualify our consensus language stating that men are not, on average, more intelligent than women, and their reliance on fringe psychologist Richard Lynn, are the same. Add to that the refusal to simply read the prior discussions where all this stuff has been adjudicated many times before, and their apparent inability to identify the sources that are clearly cited in the article. This latter point leaves me highly dubious of their claim to have only used an LLM to polish their language. I do realize that this is not a disciplinary noticeboard, and I don't think the matter yet rises to the level of WP:ANI or WP:AE, but it is relevant context. We get so much time-wasting behavior in this designated contentious topic area already, we shouldn't have our time wasted by AI too. And neither should you. Generalrelative (talk) 00:22, 15 August 2026 (UTC)
    Not knowing anything about what in the FRINGE was going on at that page and just taking this at face value, {{AINBA}} is probably warranted on account of the time wasting, should you want to try that route. ‑‑gurkubondinn 00:29, 15 August 2026 (UTC)
    Just to clarify an element of the context: I tried to discuss this directly with Generalrelative on his user talk page. I explained the arguments and source selection were my own, I read every studies I proposed, and asked that the sourcing question be discussed on its merits. That comment was removed shortly afterward with the edit “Not interested.”
    After that I narrowed the talk-page question by explicitly setting Lynn and Irwing aside and asking about the adult WAIS/ g factor literature and the sources supporting the article's categorical wording. That comment was also collapsed. I mention this because I did try to engage directly and to narrow the issue. Mill-John (talk) 11:41, 15 August 2026 (UTC)
I have collapsed Mill-John's LLM-generated post (comment link) on the fringe theories noticeboard. The formatting of the post indicates that it is more likely to be substantially LLM-generated than merely LLM-assisted. Mill-John, you are welcome to post a comment there in your own words without using an LLM. — Newslinger talk 08:30, 16 August 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

User:UnCrafteePlayZ

Latest comment: 15 days ago19 comments8 people in discussion
BLOCKED FROM MAINSPACE
Cleanup complete. (non-admin closure) M kuhner (talk) 14:39, 17 August 2026 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Green checkmarkY Cleanup has been completed, see /2026-08-17_UnCrafteePlayZ for reference.
mainblock  UnCrafteePlayZ (talk · contribs · logs · block log) (16 August 2026)
Editor has been blocked from editing the mainspace (articles only) for AI/LLM violations. 

Has been very rapidly creating new articles and expanding articles, sometimes adding articles to AIREDCATS.

Denies using AI in the edit summary of Diff/1369398452, but also expanded six stubs to full articles in an hour. ‑‑gurkubondinn 19:27, 14 August 2026 (UTC)

Just tried removing this thread in Diff/1369407842. ‑‑gurkubondinn 20:07, 14 August 2026 (UTC)
Introduced the wrong DOI for a paper in Diff/1369399320. The same paper was already cited, but this replaced the link to the paper with a DOI for some other paper instead. ‑‑gurkubondinn 20:29, 14 August 2026 (UTC)
Hello! I also noticed this user’s edits and just added the AI-generated template to a few of the pages they’ve recently expanded: [[Calcium manganate]], [[Radon difluoride]] and [[Sodium manganate]]. There are many more similar article expansions in their editing history.
I also found this diff suspicious: the edit summary says Added more CAS numbers and a few more formulas, but the diff seems to be almost entirely updating en-dashes in existing CAS numbers to hyphens.
I came to the AINB to ask what to do about this, since I’m currently using my phone’s browser and there are more articles than I can check this way. I hope it’s okay to add to this topic. I haven’t posted on any noticeboards here before, so sorry if this wasn’t the right way to do it. Suðurhafsljósæta (talk) 23:38, 14 August 2026 (UTC)
You did fine--thanks for the report!
Edit summaries which don't correspond to the edit are certainly a bad sign. (While I don't think it was LLM, I was struck by the edit I reviewed today which said "typo" in the edit summary. The edit changed "billionaire" to "businessman." Typo, huh.) M kuhner (talk) 00:30, 15 August 2026 (UTC)
Hopefully caught by the Wikipedia:Canned edit summaries filter. CMD (talk) 00:47, 15 August 2026 (UTC)
Ffs Kowal2701 (talk, contribs) 01:41, 15 August 2026 (UTC)
I also noticed large additions of dubious unreferenced text, or referenced by vague book references that probably do not include the detail referenced. Changes may be original research, but I suspect LLM output. Graeme Bartlett (talk) 11:03, 15 August 2026 (UTC)
Pretty sure it is LLM given the circular cites, red categories, and large rapid edits. ‑‑gurkubondinn 12:01, 15 August 2026 (UTC)
Americium dihydride had three refs given, of which two had invalid IDs, and one when I tracked down the journal/issue/page was an unrelated article. I draftified it. This editor needs to be given the boot. DMacks (talk) 01:37, 16 August 2026 (UTC)
Shoulda LLMPROD'ed it...will do so later tonight DMacks (talk) 01:40, 16 August 2026 (UTC)
Ok, I'm not super-familiar with that process. Given direct evidence of bad content that has LLM patterns, not just indirect presumed LLM based on editor, I think it's important to get it out of mainspae mainspace ASAP. But doesn't that set a much longer shot-clock for it, with DRAFT becoming G13 after 6 months vs PROD-llm deleted after only 5 days? DMacks (talk) 02:12, 16 August 2026 (UTC)
Ahhhh, G15. Immediate is even faster:) DMacks (talk) 02:55, 16 August 2026 (UTC)
has multiple bogus database tokens in the infobox, consisent with AI hallucination. DMacks (talk) 16:18, 16 August 2026 (UTC)

@Admins willing to patrol AINB: Editor continues to make large LLM edits and responds to attempts to communicate with reverts and hostile edit summaries. Requesting a mainspace block to force some communication. M kuhner (talk) 02:29, 16 August 2026 (UTC)

@M kuhner I blocked the user indefinitely from mainspace for "Violations of WP:EPAI, see Wikipedia:AI noticeboard#User:UnCrafteePlayZ. Refusal to communicate with other users expressing concerns about their edits." Any admin can unblock if they believe the issues raised have been satisfactorily resolved. – Epicgenius (talk) 04:11, 16 August 2026 (UTC)

Tracking page created, though we should wait a few days to see if there will be an unblock request. M kuhner (talk) 02:59, 17 August 2026 (UTC)

Most of their pages have already been dealt with! Fast work, gentlefolk. In fact there were only trivial (and correct) edits left. M kuhner (talk) 03:02, 17 August 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

User:Guywelch2000

Latest comment: 14 days ago9 comments5 people in discussion

Yet another enormous swath of AI edits going back to 2023. What makes this especially telling is that not only do their edits seem like AI, the way in which they seem like AI tracks the differences in AI model output exactly. Special:Diff/1203341299 reads like 2024 AI, Special:Diff/1288603486 reads like 2025 AI, Special:Diff/1347250351 reads like 2026 AI, etc. This also applies to their edit summaries and talk page comments, e.g. Talk:Picayune Rancheria of Chukchansi Indians, Talk:Culture of California, etc.

I warned them in February; they did not answer, despite being actively editing. Gnomingstuff (talk) 20:14, 20 June 2026 (UTC)

they responded Kowal2701 (talk, contribs) 22:36, 16 July 2026 (UTC)
from Nov 2024 had utm_source=chatgpt. Today, the site where the Sentinel Hotel once stood serves as a memory of the rich history of Yosemite Valley's development into a major tourist destination. Its enduring legacy continues to captivate those interested in the evolution of Yosemite's hospitality industry. in Sentinel Hotel (Yosemite Valley) (July 2023). I think I'm gonna draftify from Discovery Tree (Jan 2023) as it has lots of WP:AILEGACY and promo which differs from earlier creations and writing. Kowal2701 (talk, contribs) 17:34, 9 August 2026 (UTC)
I see you are moving many articles I contributed to back into Draft and out of the mainspace. I spent many hours editing and researching, especially around areas that were not well covered, including the history of the Sierra Nevada and Yosemite National Park, and these articles are well cited, and believed to have met quality guidelines for wikipedia. If tone is the issue, then could these pages be cited for tone? Guywelch2000 (talk) 17:49, 9 August 2026 (UTC)
Hi @Guywelch2000, I was about to leave a message on your talk. Normally these articles would get deleted (see WP:LLMPROD), but I can see you've been acting in good faith so I draftified them instead for you to work on if you wish (see here for a list of your drafts). The biggest issue with LLM use is hallucinations (i.e. WP:V failures), though there are other important issues like WP:NPOV/WP:WTW and WP:COPYVIO. The signs that appear as promotional writing are only how we detect LLM-generated content, copyediting is not how you fix such content (see WP:AISIGNS). Usually we recommend people to stubbify or rewrite LLM-generated articles from scratch since it's often very time-consuming to attempt to preserve such content, but it's up to you. You may want to go through WP:AFC for the first few, idk. Please let us know if you have any questions.
Unfortunately your edits to existing articles during the timeframe of LLM use will also need cleanup, can I ask if my estimate of late Jan 2023 as a start date was about right? I'll create a tracking page for this in a bit Kowal2701 (talk, contribs) 17:59, 9 August 2026 (UTC)
Thanks for the guidance and suggestions here. I have been a part of Wikipedia for about 20 years and began contributing much more often during the COVID timeline. I enjoy doing research and found that for some topics, the Chapt GPT Deep Research tool was very helpful to surface citations that were otherwise hard to locate. For example my article on Mono Winds was something that computer aided tools helped me find citations from the US Forest Service and others that I would have struggled to locate through traditional search engines. I understand what you're now saying about tone, especially neutral point of view. I did take that into account, however I have done a lot of promotional copy-writing in the past, so what sounds neutral to my ears needs to be stepped back a few more notches. Guywelch2000 (talk) 18:13, 9 August 2026 (UTC)
Can you give us an approximate start date for using LLMs to generate text? That would be very helpful in directing the cleanup. Kowal2701 suggested January 2023, is that correct? M kuhner (talk) 01:20, 10 August 2026 (UTC)
@Guywelch2000 can you give us an update? Dr vulpes (Talk) 04:29, 16 August 2026 (UTC)

I made a tracker; let's take a look at the non-creations. M kuhner (talk) 02:57, 18 August 2026 (UTC)

User:Kharbaan Ghaltaan

Latest comment: 1 month ago13 comments4 people in discussion

Kharbaan Ghaltaan (talk · contribs)

AI slop going back to at least 2024, which I first noticed in their 2024 edits on the article for Palestine (yes, certainly an article you want written with AI); I have not looked at all their edits on grounds of their being a lot of them, but through 2025 they have unambiguously used ChatGPT for at least some purposes, and had a page (User:Kharbaan Ghaltaan/Baghdad) speedied for ChatGPT usage. Also what appears to be an AI userpage.

They were asked about this last summer on their talk page as well as on this ANI thread, which points out the marked difference in their writing style between some of their talk page comments/edits and other ones (see Talk:Palestine for a sampling, apologies to use that as an example again but it's the one I just noticed and I have not gone through their enormous amount of edits to find others)

This isn't a conduct report as going through their recent edits it doesn't seem like they are using ChatGPT for article content anymore. It's mostly to flag the older stuff for cleanup, it is very frustrating to figure out where this stuff starts and ends. Gnomingstuff (talk) 00:30, 7 July 2026 (UTC)

Those older stuff is mix of AI and original written and its not that easy to filter out AI written and if that content is deleted entirely, then my entire original written content will also go in vain. Kharbaan Ghaltaan (talk) 16:58, 7 July 2026 (UTC)
You know which parts were AI and which were not. I suggest you tackle this right away. If we have to do it, you are right, some of your original writing may be lost.
If nothing else, can you give us a date when you started using AI, and a date when you stopped? M kuhner (talk) 15:31, 8 July 2026 (UTC)
There is not a proper date I remembered using AI. Just I was new and I used it and now I stopped it. The problem is, AI generated content is totally mixed with originally written ones.
Is it necessary to cleanup AI generated content, because most of them are in sandboxes and drafts, not actual articles, I guess Kharbaan Ghaltaan (talk) 07:28, 12 July 2026 (UTC)
We don't normally clean up anything but published articles and edits to published articles. I can make a cleanup report for you to look at, if you want to see what articles we would be examining. Let me know if you want to see that.
There's no way around it: if you can't tell us which articles or parts of articles are LLM, we're going to have to try to figure it out ourselves, and some of your own writing will probably get deleted. I'm sorry for that. But experience shows that LLMs can get things seriously wrong, so we can't afford to leave LLM text in Wikipedia articles. M kuhner (talk) 08:02, 12 July 2026 (UTC)
As far I remember, I didn't made any edits to published articles using LLM and whenever I did, then it was already removed. Mostly, I have used them in my draftspace and sandboxes. and also some of those articles which I wrote and published using LLMs have been either deleted, redirected or moved to draftspace. Anyway, if there is a for you to scan and inform before deleting, then I will see to it Kharbaan Ghaltaan (talk) 12:34, 12 July 2026 (UTC)
I have created a "tracking page" which lists all your edits since LLMs were publicly introduced, grouped by article. Our next step would be to go through this list and try to determine when LLM use started, then go article by article and take appropriate steps. I am not linking this page for immediate action, since you say you will help. Please have a look. If you can use this list to remember when you started using LLMRs or when you stopped, that would be extremely helpful.
Here is the tracking page: M kuhner (talk) 15:23, 13 July 2026 (UTC)
Sorry for late reply, as I have been busy nowadays. Seeing the list, one thing I am sure that most of the LLM contents were published on articles around overall Middle East topics, like countries (Iraq, Palestine etc.) and I remember, I stopped using LLMs in my India-related articles. and regarding helping, suddenly I can get busy and won't be available. Because it was all of a sudden. And please let me know that content will be removed or entire version. Because if version is not deleted, then I can rewrite those contents originally by myself. Kharbaan Ghaltaan (talk) 21:15, 19 July 2026 (UTC)
In general we delete articles that were created as LLM, and revert large LLM edits; we try not to touch anything else. M kuhner (talk) 18:25, 22 July 2026 (UTC)
regarding the edits, if the entire version is not deleted, then it will be better, because i will be able to rewrite originally that content Kharbaan Ghaltaan (talk) 09:41, 25 July 2026 (UTC)

User:Afus199620

Latest comment: 1 month ago16 comments6 people in discussion

This user has made 22,000 edits. I don't think all of them are AI, but I do know the most recent ones are. They've made five articles within minutes of each other. All show signs of AI, for instance, the first diff of this article is filled with broken refs. I've already nominated Rutunga for CSD during NPP.
First hoping for a response from the user, however looking for a consensus here on whether they should be drafted or deleted. Fortek67 (talk) 20:31, 10 July 2026 (UTC)

The articles where translations from short German articles. The text is not AI. I wont make these translations anymore.--Afus199620 (talk) 21:06, 10 July 2026 (UTC)
Nobody's stopping you from making these translations. It's just that they don't seem to be of the best quality. Are they machine translated? How exactly did you translate them/what did you use? Because once again this seems to have been done very quick. Fortek67 (talk) 21:36, 10 July 2026 (UTC)
Yes, I used a automatic translation program with editing. You are completely right. It was too fast. I will be more careful. --Afus199620 (talk) 07:10, 11 July 2026 (UTC)
You should take your own advice and check more thoroughly next time. Lupe (talk) 07:22, 11 July 2026 (UTC)
unreviewed/poorly reviewed LLM translations are still subject to the same issues as normal LLM-generated content, hallucinations and WP:V failures are the biggest problem Kowal2701 (talk, contribs) 10:01, 11 July 2026 (UTC)
I know that you have to review sources of machine translated articles and that that takes maybe a minute or so longer even for these short articles, but that wasn't the accusation (it wasn't even mentioned that the articles are translations). So I stand by my statement for him to work on himself. Lupe (talk) 11:39, 11 July 2026 (UTC)
I don't see that Fortek said anything wrong here? The articles were AI-generated and problematic, it's just that we allow AI-generated translations that are reviewed. Though I agree saying that they're translations in the OP would be ideal Kowal2701 (talk, contribs) 11:46, 11 July 2026 (UTC)
Not just ideal but required. The original source needs to be credited. See WP:Translation. M kuhner (talk) 15:56, 11 July 2026 (UTC)
(I was referring to Fortek's OP here, but yeah, unattributed translations are a pain) Kowal2701 (talk, contribs) 16:03, 11 July 2026 (UTC)
I'm sorry, I wasn't clear: what I was trying to say is that if these are translations, they are unattributed translations, which is also a policy violation. M kuhner (talk) 16:07, 11 July 2026 (UTC)
The Translations where actually not unattributed. The revision history showed that these articles were translations.--Afus199620 (talk) 17:44, 11 July 2026 (UTC)
You are right. I apologize and will strike out my remark. M kuhner (talk) 17:48, 11 July 2026 (UTC)

(The following has been copied from a separate case, with PARAKANYAA's permission, as they turned out to be about the same editor.) M kuhner (talk) 03:47, 26 July 2026 (UTC)

This is more subtle and I'm not sure if I'm being paranoid but I'd like a second opinion. My concerns are about these edits to Peter Levenda I removed a chunk of it for being undue weight on the author's self-description /borderline copyvio but reading further I think AI might have been involved. Not sure though. My reasoning:

1) a lot of the analysis of the books is not in the sources and reads, to me, like AI extrapolating the contents from more dry summaries in reviews. The above section is just basically copying from the author summary (which also seems like copyvio ) so this mostly shows up in the views section

2) phrasing like Levenda's books generally approach occultism as a historical, political and religious phenomenon rather than as a purely literary subject. Unholy Alliance examines the relationship between Nazi ideology, esotericism and occult currents, while later works such as Sinister Forces and The Dark Lord connect occult history with American political culture, H. P. Lovecraft and modern esoteric movements. In the Sekret Machines series, written with Tom DeLonge, Levenda applied similar themes to the modern UFO and UAP discourse, arguing that the subject should be considered in relation to religion, mythology, consciousness and political authority rather than only as a question of aerospace technology and was described in contemporary press coverage as being based on interviews with intelligence officers, scientists, engineers and military officials really smell of AI to me.

3) repeating the same information across multiple redundant sections, e.g. it covers the UFO books twice? PARAKANYAA (talk) 17:51, 25 July 2026 (UTC)

Oh, I notice this user is already being discussed above (Afus199620). My bad. Still, the problem here is a bit broader than what is discussed there, the problem goes beyond translations. PARAKANYAA (talk) 17:57, 25 July 2026 (UTC)

User:Gamji231

Latest comment: 22 days ago2 comments2 people in discussion

Gamji231 (talk · contribs) XTools

see this edit, it links a paper named "Preferences for Cash vs. Digital Payments" that has nothing to do with a Harley-Davinson advertisement featuring a hijab, says " in place of the brand's traditional leather-clad rider imagery" when the cited source talks about an ad that shows half the face of one woman in a hijab and half of another in leather, has an overwhelmingly exhaustive editing summary, and this textbook WP:AICOMMENT featuring oaicite

had a warning for using LLMs in february ~2026-38678-64 (talk) 18:38, 13 July 2026 (UTC)

Some signs they gamed ECP with wikilink addition (a ton of those now reverted as useless/harmful). The edit summaries change over time but are never reassuring (they like to say "A small edit" as an edit summary and then add a sizeable chunk of text). I'd be inclined to say LLM suspicions go all the way back to the start (July 2025), but I haven't found conclusive evidence. (Edit summary "Added a new section with credible sources from reputable media outlets." does not make me happy but it's not a smoking gun either.) By November 2025 I'm sure: Added “Miss Universe 2025” section with verified sources and internal links. Expanded coverage to include Nadeen Ayoub’s participation as the first Miss Palestine delegate, international media reporting, and contextual significance. Maintained neutral tone per WP:NPOV and cited multiple reliable sources (The New Arab, Times of Israel, Israel Hayom, AS, Grazia ME, NY Post, The Sun, The Times, The Canary).. Making a tracker, though the wikilink gaming is going to clutter it badly. M kuhner (talk) 03:40, 10 August 2026 (UTC)

BlueTigerBlue00

Latest comment: 1 month ago8 comments5 people in discussion

Information icon There is currently a discussion at Wikipedia:AI noticeboard regarding an AI cleanup investigation which may be related to you.

warned  BlueTigerBlue00 (talk · contribs · logs · block log) (25 July 2026)
Editor has been warned about the use of AI/LLM content and WP:NOLLM. 

BlueTigerBlue00 (talk · contribs)

LLM, exhaustive summary, here and there ~2026-38678-64 (talk) 21:37, 13 July 2026 (UTC)

The second edit (Diff/1351359381) adds a source with an utm_source=chatgpt.com url parameter. Not necessarily a smoking gun in and of itself, but probably worth looking into. ‑‑gurkubondinn 21:44, 13 July 2026 (UTC)
i was alerted to their edits by mistakenly thinking that they added referrer=grok to the article, it seems that i have to file another AINB for that sockpuppet (eugh) ~2026-38678-64 (talk) 21:47, 13 July 2026 (UTC)
They responded with

Thanks for the note, but thats not correct. I write all content myself, based on reliable sources. The only tool I use sometimes is an AI spellchecker to correct my grammar and typos before posting, because I'm native German but mostly do work on Engl wiki. The text itself is entirely my own words, not LLM generated BlueTigerBlue00 (talk) 00:07, 18 July 2026 (UTC)

Re this edit (23 July), the source doesn't even mention "Seen", nor does it mention commercialisation (also WP:IMDB). @Admins willing to patrol AINB: , I think a mainspace block is needed here Kowal2701 (talk, contribs) 11:02, 24 July 2026 (UTC)
Hi Kowal2701, Seen and the film details are mentioned in the cited IMDb page (cited archive link), so I can't conclude LLM use from that, despite IMDb being an unreliable user-generated source. However, their explanation does not address their addition of a ChatGPT-generated link. — Newslinger talk 17:54, 25 July 2026 (UTC)
Sorry, I need to be checking archived links. Will keep an eye on this then, thanks Kowal2701 (talk, contribs) 20:08, 25 July 2026 (UTC)
All, or nearly all, of their big content edits have an "Edit Check (paste) shown" tag, which means they're copying and pasting from somewhere into Wikipedia. They could be copying from AI, or they could be copying from their spellchecker program. ARandomName123 (talk)Ping me! 01:42, 27 July 2026 (UTC)
Oh great, they removed an LLMPROD saying they'd reviewed it all , the first ref I checked (ref #8 ) doesn't mention several of those guest stars. Kowal2701 (talk, contribs) 14:59, 31 July 2026 (UTC)

Ivanhardybirt

Latest comment: 22 days ago4 comments1 person in discussion

Ivanhardybirt (talk · contribs), XTools

Their first LLM article appears to have been on 12 April 2025 (stark difference to their earlier creations, obvious signs), they've created 26 since with some deleted. Was brought to ANI October 2025 (at which time they were warned for LLM use ) which got archived as unresolved. Their most recent creation was 11 July Kowal2701 (talk, contribs) 22:42, 14 July 2026 (UTC)

well, that is upsetting Kowal2701 (talk, contribs) 17:43, 17 July 2026 (UTC)
I've draftifed creations that weren't obviously human-written from 12 April 2025 onwards, but I'm not convinced that's all of them as their use of AI seemed sporadic Kowal2701 (talk, contribs) 21:16, 9 August 2026 (UTC)
noting User talk:Ivanhardybirt#LLM use, nice to see Kowal2701 (talk, contribs) 18:36, 10 August 2026 (UTC)
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:AI_noticeboard/Archive_13&oldid=1370002431"