Wikipedia talk:WikiProject AI Cleanup

Main pageDiscussionGuide and resourcesPoliciesResearch

Can someone check this discussion?

This discussion (to me) looks to have some LLM responses in it: Talk:Battle of Lumë#RFC on Infobox for Battle of Lumë. I don't want to collapse/ask myself because I've !voted in the RfC. InfernoHues (talk) 20:53, 25 July 2026 (UTC)

I skimmed it but didn’t see signs beyond the length. Any specific passages? Dw31415 (talk) 23:48, 25 July 2026 (UTC)

Some of FranéRogoz and Magapetro's comments seemed weird to me. Especially how they seem to kind of but not really respond to what people said. InfernoHues (talk) 02:43, 27 July 2026 (UTC)

Pages by project

I thought I'd get to work on some namespage pages after all my blathering at the guideline. I tried one at Talk:Abas (mythology)#Topos Text is AI? and was thankfully rescued by another editor. It dawned on me that it'd be much easier to do on pages I care about.

I created User:Dw31415/AI Cleanup to contain the projects I care about. It was a good exercise to continue to learn the db schema so I didn't spend much time looking to see if this was already done.

Probably the best place to put this is on a subpage of each project I'm interested in (like Wikipedia:WikiProject Medicine/Deprecated archival service).

Any thoughts, recommendations, requests?

Article	Project	Rating	Importance
1965 British European Airways Vickers Vanguard crash	Aviation	Start
1968 Sainte-Marie Douglas DC-6 crash	Aviation	Start

Thanks to all you editors on the this project! Full disclosure, I created the sql using an agent. Dw31415 (talk) 12:27, 26 July 2026 (UTC)

@Dw31415: Nice. I have no idea how these work, but would it be possible to create a template so that interested WikiProjects can just simply substitute that and create a page like this? Kovcszaln6 (talk) 13:18, 26 July 2026 (UTC)

The projects go right at the top so it’s pretty easy to modify (see below). I like the idea of the template but won’t have time to do that for another week or so. I’d be happy if someone else wants to do that. It’d be helpful to have one even without parameters because it’s not obvious where the sql template starts and the results table ends.

WITH selected_projects AS (

/* Edit this list to choose projects. Titles are PageAssessments titles. */

SELECT 'Aviation' AS project_title

UNION ALL

SELECT 'Film' Dw31415 (talk) 14:15, 26 July 2026 (UTC)

maybe Template:AI Tag By Project is a good name? Dw31415 (talk) 22:11, 26 July 2026 (UTC)

User scripts for AI image detection + marking

AIMarker adding an AI-label to generated images

Hi all, I've got a couple user scripts for anyone interested in identifying AI media.

• AIMarker adds an AI watermark to images that have been identified as having been AI generated, based on the templates and categories on the filepage. This doesn't reveal anything new, but more readily surfaces the fact that an image has been AI generated or modified.
• Over on Commons, AIDetect lets you search a file's metadata for signals of AI generation (e.g. C2PA statements, standard diffusion metadata, etc.). This can be helpful to find confirmation that an image has been AI-generated/modified (but of course it's easy for metadata to be stripped from an image by screenshotting, downloading from social media, etc., so it's only as good as the metadata that's present).

Thanks, ~SuperHamster Talk Contribs 13:15, 27 July 2026 (UTC)

AI asked to describe a work of art instead describes the infobox image on the artist's Wikipedia page

Self-portrait (Yayoi Kusama) - this is a new one for me. I was like "why does this article describe an acrylic painting as a digital photograph?" Then I looked at the infobox image on Yayoi Kusama and it all became clear. -- LWG ^{talk (VOPOV)} 18:32, 27 July 2026 (UTC)

Is there a truer self-portrait than the face that stares back at you as you look at your Wikipedia page? CMD (talk) 00:38, 28 July 2026 (UTC)

The Simple Summaries did this sometimes, couple examples: A cuckold is a term used to describe a husband whose wife is unfaithful. It comes from a 19th-century painting by Cornelius Krieghoff., or This ancient painting shows Hercules carrying his son while facing the centaur Nessus. Hercules is a famous Greek hero, the son of the god Zeus. He's known for his strength and bravery, fighting monsters and completing tough tasks. In Roman times, he was called Hercules and was very popular. This picture tells a story from his adventures Gnomingstuff (talk) 20:15, 28 July 2026 (UTC)

The first version was somehow even worse. I've tagged the article for G15 because other than the issue you pointed out, there is communication intended for the user present in the infobox. Athanelar (talk) 12:15, 30 July 2026 (UTC)

Discussion at WT:AISIGNS § New AIS: shortcut format

 You are invited to join the discussion at WT:AISIGNS § New AIS: shortcut format. Athanelar (talk) 12:11, 30 July 2026 (UTC)

AI noticeboard

I'm aware there is a banner saying to report AI content on the AI noticeboard. I couldn't find instructions for how to list a single article. I have no intention of engaging in any cleanup or discussion of this article but Al Sa'dun is clearly LLM made and has some issues. If there is another process I should go through to report this, let me know and I'll do it, but I just wanted to inform the right people and stop thinking about it. I would recommend to the project to create instructions for the AI noticeboard for someone less familiar with the systems in place, but y'all probably have more than enough on your plate as is. – Ike Lek (talk) 20:09, 30 July 2026 (UTC)

Hi -- for something like this you can add the AI-generated tag to the top of the article and list the reason in the parameter, for instance, {{AI-generated|date=July 2026|reason=whatever your reason is}}. You will probably want to add this reason text to the talk page as well. Gnomingstuff (talk) 22:28, 30 July 2026 (UTC)

It's worth looking at the user who created it to see if there's much more to cleanup, in this case it's AytWaygher232 (talk · contribs) who's also created which appears to have a hallucinated category (WP:AIREDCAT). I've warned them but haven't done any cleanup Kowal2701 (talk, contribs) 17:03, 31 July 2026 (UTC)

Yeah, I think this might be someone who has read Wikipedia:Signs of AI writing and is trying to avoid the major signs. The red cat was a good catch since no category exists for any of the alternative spellings of the place, making a typo unlikely. Even then, it would be easy to argue that they were planning on creating the category. I can tell with a pretty high level of confidence that LLM generated text was used heavily in the article, but it is difficult to put together a case to prove it. I find this kind of work frustrating to a point that I choose not to do it most of the time. I respect what this project does, but I believe I can do a lot more good for Wikipedia as a whole if I don't burnout doing something I hate. Ike Lek (talk) 17:33, 31 July 2026 (UTC)

Probably more likely to be just an AI translation -- translated material can sometimes deviate more than non-translated text from the standard set of AI signs since there's at least some original text to influence the final product.

(but yeah, for every article I tag there are two or three I don't because, as you say, it is difficult to explain) Gnomingstuff (talk) 05:03, 1 August 2026 (UTC)

You can also just file a one line report to the noticeboard, there is no requirement to even inform the user. LLM-generated "translations" are just as much LLM-generated text as non-translation output is, and are a constant source of problems for the project.

For the record, I hate doing the cleanup too. Most of the time, I try to stick to identification (and all of the associated communication), been trying to get back some wikitime to do things that I actually enjoy around here. ‑‑gurkubondinn 12:09, 2 August 2026 (UTC)

Good to know. New account is on the scene removing the tag. I put it back. If they create more trouble, I can try making a report there. @Kowal2701 you seem to have a good understanding of this. Would you be willing to keep an eye on the page and see if more newly created accounts pop up adding large chunks of text? Starting in a few days, there's a high chance I'm going to be a lot less active for a couple weeks while I deal with some offline stuff, but I don't want this to be forgotten about. – Ike Lek (talk) 00:47, 3 August 2026 (UTC)

I can put it on my watchlist (btw I have a much looser grip on things than the other two people here lol, that made me chuckle) Kowal2701 (talk, contribs) 01:07, 3 August 2026 (UTC)

Thanks, I appreciate it, and I meant absolutely no insult to the other people here. You do important and difficult work. Ike Lek (talk) 01:27, 3 August 2026 (UTC)

The only thing Kowal is wrong about is "having a looser grip". And just for future reference, if you make a (however short) post on the noticeboard then it probably won't be forgotten about, because Kowal regularly goes through the noticeboard archives and compiles lists of stuff that needs to be looked at. ‑‑gurkubondinn 07:35, 3 August 2026 (UTC)

User generating incomplete templates and strange summaries - possible AI usage?

See: Special:Contributions/Holuwadamilarecomics and edits such as: Category:User navigation templates and Template:User likes books. Also this unregistered edit which I believe is the same person: Special:Contributions/~2026-42799-37 - Wil540 art (talk) 13:42, 3 August 2026 (UTC)

Discussion at Wikipedia talk:WikiProject Insects § AI-generated images?

 You are invited to join the discussion at Wikipedia talk:WikiProject Insects § AI-generated images?, which is within the scope of this WikiProject. Ham II (talk) 18:59, 3 August 2026 (UTC)

Rewriting the guide

Hi, I've rewritten the guide and combined it with the resources page, see Wikipedia:WikiProject AI Cleanup/Guide and resources (haven't swapped them out yet). It's mostly aimed at onboarding people to AINB, but is mostly me rambling on and pretty much just represents my views on stuff so please feel free to boldly edit it. The patrolling section needs fleshing out a bit, I wasn't sure what advice to give people. Feedback is very welcome Kowal2701 (talk, contribs) 19:09, 3 August 2026 (UTC)

Maybe something about cleanup for smaller cases, the ones where someone adds 1-2 paragraphs? Those are a lot less daunting and easier to salvage than whole articles.

What I usually do in those cases, if I'm not outright deleting the text, is to "stubify" it by extracting the basic facts. For example, someone added a paragraph of slop regarding an album a musician released, I'd turn that into "In 2026, Artist released Album" (assuming that was accurate and sourced, obviously). Gnomingstuff (talk) 20:14, 4 August 2026 (UTC)

thanks, I've put For some additions, or for when the prior version of the article is of poor quality, it's worth considering salvaging basic facts from the content., I'll probably add it today if there are no objections Kowal2701 (talk, contribs) 14:55, 8 August 2026 (UTC)

done Kowal2701 (talk, contribs) 18:51, 8 August 2026 (UTC)

The color of the collapse containing the answer spoils the answer itself

Title. Please make them all the same color (be it green or gray). - ~2026-42525-36 (talk) 17:29, 6 August 2026 (UTC)

Oh I forgot that not all pages are edit-protected. I'll go do that myself. - ~2026-42525-36 (talk) 17:30, 6 August 2026 (UTC)

Warn templates for creating article

Hi, does anyone fancy rewriting {{uw-create-ai1}}, {{uw-create-ai2}}, and {{uw-create-ai3}}? They still have pre-NEWLLM versions Kowal2701 (talk, contribs) 00:38, 10 August 2026 (UTC)

I started doing this, but these templates are anyway obsolete now considering there's no longer a distinction between creating articles with AI and just plain editing with AI, so I'll tag them as {{historical}}. Athanelar (talk) 07:20, 12 August 2026 (UTC)

@Kowal2701 I've tagged these as historical and also modified them so that when substituted/transcluded they show the following warning (or variant thereof) instead of their text;

I couldn't get the -1 template to cooperate with this so I removed the original template text on that one, maybe somebody can fix it. Athanelar (talk) 07:43, 12 August 2026 (UTC)

Help with cleanup

I encountered a case where an editor used AI-generated text to make major changes to several articles over years, see the explanation at User_talk:Thiagovscoelho#Query.

I went through their most recent 500 edits, going back to 4 November 2025, focusing on big changes. For articles where there were no major changes after their edits, I reverted back to the version before the edits. For articles with major changes afterwards and for articles that they created, I added the maintenance tag "AI-generated".

It might be good if someone with more experience on AI cleanup could have a look at this problem and assess what the best response is. Edits before 4 November 2025 probably have the same issues, so there may be more to do, but I'm not sure how far back to go. In case someone wants to continue, here is the edit history before 4 November 2025. Phlsph7 (talk) 13:43, 10 August 2026 (UTC)

Could you post this on WP:AIN? ‑‑gurkubondinn 22:13, 10 August 2026 (UTC)

Done, see Wikipedia:AI_noticeboard#User:Thiagovscoelho. Phlsph7 (talk) 08:10, 11 August 2026 (UTC)

Upcoming Anthropic/Claude text watermarks

Just a heads up for project members: Anthropic will be adding embedded watermarks to generated text, and providing some sort of tool to allow people to check if text carries the watermark. I assume it will be some sort of statistical pattern in the choice of words, similar to Google's SynthID-Text. ~SuperHamster Talk Contribs 02:32, 11 August 2026 (UTC)

That's brilliant news (I was worried they'd avoid doing this and lean on diplomatic support from the US gov), it has to be in place in all models by 2 December per the EU's AI act Kowal2701 (talk, contribs) 11:22, 11 August 2026 (UTC)

And now an FAQ on watermarking from Anthropic. ~SuperHamster Talk Contribs 22:28, 14 August 2026 (UTC)

so it's like SynthID, doesn't look like we'll be able to automate it Kowal2701 (talk, contribs) 22:53, 14 August 2026 (UTC)

Well, they'll have an API to check text against. Depending on the access restrictions, could be something we build tools with. ~SuperHamster Talk Contribs 23:07, 14 August 2026 (UTC)

• Jason Koebler (18 August 2026). "Anthropic's Text Watermarking Proves AI Companies Do Not Care at All About Writing". 404 Media. Archived from the original on 18 August 2026.

‑‑gurkubondinn 10:21, 24 August 2026 (UTC)

Some important info regarding AI as a source-search tool

I've stumbled upon some information from various studies suggesting that AI tools with web search functionality/retrieval-augmented generation (i.e., most of them now) are essentially being penned in to a limited subsection of the internet where they are statistically more likely to be pulling information from misinformative sources, or sources also generated by AI; because mainstream news and information outlets are blocking AI webcrawlers. The implication for Wikipedia is that if editors are using AI as a tool to find sources for their articles (which we currently promote as a responsible use of AI) they are seemingly statistically likely to be being fed misinformation or AI slop.

Analyzing a curated dataset, we find a stark contrast: 60.0% of reputable sites disallow at least one AI crawler, compared to just 9.1% of misinformation sites...Finally, our longitudinal analysis reveals that this gap has widened over time, with AI-blocking by reputable sites rising from 23% in September 2023 to nearly 60% by May 2025.

Our findings show evidence of AI-generated sources being cited across all four generative search engines [ChatGPT, Copilot, Gemini, Perplexity] (~16% of cited sources)

I've also found a couple of sources like this one claiming that listicles are the most commonly-cited type of source by these AI tools. The source linked there claims that out of 75,000 answers, 21.9% of the citations were to listicles.

In summary, editors who are using AI to find sources (which we promote as a responsible use, as I said) are being subtly funnelled away from reliable sources and into a low-quality information ecosystem full of AI slop, listicles and outright misinformation. I think we therefore need to update our guidance on this topic and encourage people not to use AI for source searching. Athanelar (talk) 07:12, 12 August 2026 (UTC)

Before this has any chance of success, we'd need to be able to offer concrete alternatives. Is it possible to get a genuinely non-LLM search out of Google, for example? Or are they all LLM, but differently packaged? M kuhner (talk) 18:26, 12 August 2026 (UTC)

I've used LLMs to find sources before, and thought that they worked decently well. I normally tell it to "provide academic sources on X". That said, it still missed some high quality sources I know existed that were better than the provided sources. InfernoHues (talk) 18:31, 12 August 2026 (UTC)

This tracks with my gut sense of what I've seen in AI output -- works cited usually favors either official self-published websites or stuff like Forbes etc. -- but the data I have is very skewed because AI-generated text is disproportionately likely to also be intended as promotional and/or refspam, which affects the token distribution, and there's no easy way to correct for that.

For instance, comparing the AI post August 2025 dataset to the dataset of human articles tagged promotional/peacock/userspace draft, disproportionately common tokens are stuff like "|title=Introducing", "|title=Our", "|title=Official", "|publisher=City", "|publisher=Ministry", but even with the restriction on human text there are lots of potential explanations other than it being AI. Gnomingstuff (talk) 19:05, 12 August 2026 (UTC)

Lumberjack essay about AI proofreading

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) 16:07, 12 August 2026 (UTC)

I don't recall that essay, but I do now want to rewrite The Lumberjack Song to "He's a lumberjack and he's AI"! Thryduulf (talk) 16:15, 12 August 2026 (UTC)

Do it! Do it! M kuhner (talk) 18:21, 12 August 2026 (UTC)

I remember this -- it was a table of possible edits to a sentence and subtle meaning changes introduced -- and remember contributing to it, but couldn't find it in my contributions, maybe it was deleted? Gnomingstuff (talk) 18:37, 12 August 2026 (UTC)

If you can give me a hint of what you might have written in your edit summary, and a rough date range, I can look through your deleted contributions and restore it somewhere, like user space or draft space. ~Anachronist (who / me) (talk) 15:08, 13 August 2026 (UTC)

I know for certain that I first encountered the page this year, probably May–July 2026. –LaundryPizza03 (dc̄) 17:37, 13 August 2026 (UTC)

Your deleted contributions have about 2000 edits in the May-July range, more than 1000 of which are in May alone. If you have any clue that would narrow it down, let me know. ~Anachronist (who / me) (talk) 23:35, 13 August 2026 (UTC)

@Gnomingstuff –LaundryPizza03 (dc̄) 14:20, 14 August 2026 (UTC)

not sure why I was pinged to this, I have nothing more to contribute Gnomingstuff (talk) 14:48, 14 August 2026 (UTC)

Because you surely remember something that can be used to identify the page.

@Anachronist: In this case, since Gnomingstuff claims to know nothing, I'd suggest doing a brute-force search. –LaundryPizza03 (dc̄) 15:28, 14 August 2026 (UTC)

I tried searching for obvious strings like "lumberjack" or "LLM" or "proofreading" and others, and nothing came up in those 2000 edits. Do you remember if it was in someone's user space, or Wikipedia namespace? ~Anachronist (who / me) (talk) 17:34, 14 August 2026 (UTC)

No. I'd lean toward checking Wikipedia namespace first. Keyword might be revision (as an uncountable noun) or proofreading. –LaundryPizza03 (dc̄) 17:38, 14 August 2026 (UTC)

No luck. You hardly have any edits in the Wikipedia namespace (Wikipedia talk has much more). Here's what you have for the past 2 years:

• 00:06, 13 August 2026 Wikipedia:Ahora en TODOS los Hipermercados Carrefour Bajamos los precios Requesting speedy deletion (CSD G3, CSD G11).
• 00:05, 13 August 2026 Wikipedia:¡ENCONTRÁ EL REGALO IDEAL PARA EL DÍA DE LA NIÑEZ EN CARREFOUR! Requesting speedy deletion (CSD G3, CSD G11).
• 00:03, 14 May 2026 Wikipedia:Sockpuppet investigations/Daisuke Roppunganai Requesting speedy deletion (CSD G6, CSD G7).
• 00:02, 14 May 2026 −999 Wikipedia:Sockpuppet investigations/Daisuke Roppunganai Oops, misspelled Tag: Blanking
• 00:01, 14 May 2026 N Wikipedia:Sockpuppet investigations/Daisuke Roppunganai Adding new report for Daisuke Roppunganai.
• 04:31, 13 February 2026 Wikipedia:建立条目向导 Requesting speedy deletion (CSD G15).
• 02:43, 5 January 2026 Wikipedia:WikiProject GuCe Editors/Balog elimination drives Requesting speedy deletion (CSD G1).
• 19:22, 31 December 2025 Wikipedia:Articles for creation/Draft Requesting speedy deletion (CSD G2).
• 07:49, 17 December 2025 Wikipedia:Cameron Jay Hauke Requesting speedy deletion (CSD G3).
• 07:46, 17 December 2025 Wikipedia:Luzhou Weing Lei Bu Requesting speedy deletion (CSD G3).

Everything prior to that is 2022 and earlier. ~Anachronist (who / me) (talk) 17:46, 14 August 2026 (UTC)

I have never edited the page you're looking for. I assumed, up to this point, that you were looking through Gnomingstuff's contribs, not mine. –LaundryPizza03 (dc̄) 17:48, 14 August 2026 (UTC)

Unfortunately, Gnomingstuff has zero deleted contributions in the Wikipedia namespace, and the deleted contributions in user space all have an edit summary "u6 triage per policy discussion...". ~Anachronist (who / me) (talk) 17:56, 14 August 2026 (UTC)

Is it possible that Gnomingstuff edited the page further back? It is also likely that if it was in Wikipedia namespace, there would be a revision to Wikipedia:Artificial_intelligence_resources that removed a link to the deleted entry. –LaundryPizza03 (dc̄) 18:05, 14 August 2026 (UTC)

Okay, not that page, but possibly a different page in Category:Wikipedia and artificial intelligence or the WP:AINB archives linked to it. –LaundryPizza03 (dc̄) 18:06, 14 August 2026 (UTC)

The lists on category pages are auto-populated without any changes to the revision history.

Gnomingstuff has zero deleted edits in the Wikipedia namespace for all time, not just this year. ~Anachronist (who / me) (talk) 20:41, 14 August 2026 (UTC)

That second remark referred to the page histories of other pages in Category:Wikipedia and artificial intelligence, not the category itself. Maybe someone else familiar with this page can chime in. –LaundryPizza03 (dc̄) 00:01, 15 August 2026 (UTC)

Proposed AISIGN for periphrastic connections

 You are invited to join the discussion at Wikipedia talk:Signs of AI writing § Connections. — EmoryNB ^{(they/them) talk contribs} 16:59, 14 August 2026 (UTC)

More WMF AI slop incoming

https://www.mediawiki.org/wiki/VisualEditor/Suggestion_Mode/Model-generated_editing_suggestions

Because the AI edits we get from Newcomer Tasks are SO AMAZING that now we are making them even easier Gnomingstuff (talk) 06:30, 15 August 2026 (UTC)

These are, apparently, the list of suggestions. They made this really difficult to find, for some reason Gnomingstuff (talk) 06:37, 15 August 2026 (UTC)

Skimming/spot-checking, most of these are fairly obvious band-aid fixes that, if you needed a tool to point them out, you probably shouldn't be copyediting articles. The suggestion text is obviously slop that aims to ensure and enhance the streamlining and aligning with etc. Unsurprisingly, the "NPOV" suggestions tend to be the worst.

Sometimes the suggestions have already been fixed:

• Bomb-making instructions on the Internet: Was (very obviously) vandalized in Special:Diff/1358132635, and the vandalism got reverted by ClueBot basically immediately. The suggestion nevertheless refers to the vandalized version (Remove the non‑encyclopedic, opinionated rant). I don't know how this happened.

And some suggestions are just actively bad:

• Frost Children, regarding a sentence about the album called Smile! :D: The sentence contains an emoticon, which is non‑neutral and informal. If someone followed this suggestion they would turn a correct statement into a wrong one. (See also 1797 in Denmark: The word "pyusician" is a misspelling; it should be corrected to "musician". The actual correct spelling is "physician".)
• Enzo Cilenti: This nonsense: The phrase ""girlfriend"" is outdated and introduces a personal tone that is not encyclopedic; it should be replaced with a neutral description of their professional collaboration and marital status.. This may come as a surprise to some people, but girlfriends exist. They are a real category of person that exists in the world. The word is no more or less neutral than "wife" or "partner." (I invite anyone reading this who does have a girlfriend to tell her she's not your girlfriend because it's outdated.)
• Vimbuza: Was flagged The phrase “a testament to the resilience and strength of this ancient tradition” is value‑laden and should be removed to maintain neutrality. Which is not exactly wrong, but the reason it is "value-laden" is because the article is AI-generated (created by a serial LLM user), and following this suggestion would amount to putting a bandaid on the problem. (This seems to happen semi-frequently; a suggestion in Jeremy Coller points to the AIATTR in Special:Diff/1326330155, but does not identify the actual problem.)
• FreedomWorks: Did you think we'd get through this without some political or geopolitical disaster? The original wording presents FreedomWorks' actions as definitively false and misleading, which is non‑neutral. The sentence is cited to a Washington Post article that indeed names several unambiguously false and misleading claims made by FreedomWorks. Did Elon Musk hijack the LLM?

Gnomingstuff (talk) 07:12, 15 August 2026 (UTC)

Wow. Yeah OK. "Rewriting language in a neutral point of view" (if NPOV is as defined by WP:NPOV, i.e., specifically relative to what is published by RS) is certainly not something I would consider "low risk" or something I'd expect a generic model from a year ago to be capable of at anything approaching an acceptable level of quality. Alpha3031 (t • c) 13:16, 15 August 2026 (UTC)

other miscellaneous weirdness:

• For some reason NPOV is described as "Re-writing sentences based on a natural point of view"; I don't know whether that's surfaced or internal
• The LLM seems to have problems interpreting wikitables and will often intepret them as if the article itself contains the JSON snippet they received: Remove the nonsensical JSON list and replace it with a brief, readable statement or omit it entirely. (Markazi Jamiat Ahle Hadith). Imagine getting this suggestion as a new editor.
• The suggestion itself is fine but this is just hilarious: The word 'ithe' is a typo; it should be 'the'. Fixing this error improves the sentence's correctness.

Gnomingstuff (talk) 07:31, 15 August 2026 (UTC)

The first article on that list, History of Botswana, is a high quality article written by a quite decent editor. I do not find the NPOV objection the csv makes compelling. The second article, Waco F series, raises a similar supposed NPOV objection about popularity. The third NPOV objection, to Hong Kong and Macao Affairs Office, does seem like a sentence that could be improved due to a misreading of the source, but I wouldn't class it as necessarily a very obvious NPOV issue. CMD (talk) 07:47, 15 August 2026 (UTC)

Some more of the many ways this can go and has gone wrong.

Just like last time, the LLM sometimes treats captions, sidebars and other incidental text as if it is article prose, and makes "suggestions" accordingly:

• 2019–20 Philadelphia 76ers season: The lead repeats ""NBA professional basketball team season"" twice, creating redundancy. Obviously, if you go to the actual article, it does not; the "sentence" the LLM received was "NBA professional basketball team season NBA professional basketball team season The 2019–20 Philadelphia 76ers season was the 71st season of the franchise in the National Basketball Association (NBA)."
• Chen Lijun (actress): Redundant repetition of the subject’s occupation; the phrase “Chinese” appears twice. The first "Chinese" comes from the "lang-zh" template.

The thing seems to go sentence-by-sentence, or at least tries to, which means it gets it wrong in some cases:

• Blockquotes as in Mrs. Miniver: The statement ""Joseph Goebbels, minister of Nazi propaganda, wrote:"" is incomplete and lacks supporting content, violating neutral point‑of‑view standards. Obviously, there is a quotation below this, and preceding a quotation with a fragment like this is acceptable per MOS:BLOCKQUOTE. Not even going to touch the "violating neutral-point-of-view standards" bit.
• Multiple short sentences sometimes seem to be treated as one, as in "While sailing aboard Masquerade, Olivia proposes to Tim, but he is reluctant, saying he was previously convicted for writing bad checks. He also admits his affair with Brooke. His "honesty" impresses Olivia, and the couple marry." from Masquerade (1988 film). This produces the nonsensical suggestion The sentence combines multiple ideas about Olivia's proposal, Tim's reluctance, his past conviction, his affair, and their marriage, making it overly long.

Some stuff is just perplexing:

• 1989 Indian general election in Jammu and Kashmir: The phrase "Jammu and Kashmir National Conference won 3 seats, Indian National Congress won 2 seats, and an Independent candidate won 1 seat" uses mixed numeral styles and can be simplified. I have no idea what this is talking about or how the sentence could be made any simpler without pointlessly chopping it up.

On a broader level, I just do not know who this is for. Experienced copyeditors don't need this. Non-fluent English speakers won't be able to do anything with any of this beyond the most simple typo fixes. The fragile baby bird newcomer that apparently exists in the imagination of whoever keeps proposing these features will probably just be confused by the lack of concrete changes suggested. (Not that adding actual AI-generated suggestions would be an improvement.) Gnomingstuff (talk) 16:06, 15 August 2026 (UTC)

What a #$%&ing nightmare. Did the model think the Jammu and Kashmir election results were meant to be read phonetically, resulting in it parsing "one 3 seats, one 2 seats, one 1 seat"?? Sarsenet•he/they•(talk) 17:31, 15 August 2026 (UTC)

More eyes on articles by User:Kadesthetics please

It appears that @Kadesthetics has been extensively adding material to astronomy articles using AI, and more recently created one. I believe these have now all been reverted or in one case draftified. However there are some older article additions in areas I am less familiar with that probably should also be checked. Note that they were warned twice by others a few days ago, and hopefully this was just a newbie mistake. Ldm1954 (talk) 02:23, 16 August 2026 (UTC)

Discussion at WP:VPI § Watchlist notice for WP:MAINTENANCE

 You are invited to join the discussion at WP:VPI § Watchlist notice for WP:MAINTENANCE. Kowal2701 (talk, contribs) 21:05, 16 August 2026 (UTC)

The size of AIN is out of hand

Moved to WT:AIN § Noticeboard size is out of hand

 – ‑‑gurkubondinn 13:55, 18 August 2026 (UTC)

Discussion at Template talk:Talk header#Add note about WP:AITALK

 You are invited to join the discussion at Template talk:Talk header § Add note about WP:AITALK. Dreamyshade (talk) 05:32, 20 August 2026 (UTC)

WP:GRAMMARLY currently seems unhelpful

 You are invited to join the discussion at Wikipedia talk:Don't use Grammarly § I don't think Grammarly users will get the point. – MrPersonHumanGuy (talk) 17:00, 24 August 2026 (UTC)