Wikipedia talk:WikiProject Articles for creation

It looks like we've created a new defacto quick fail for LLM content. The apparent justification is WP:NEWLLM. I don't know if this has been discussed here but it has not made it to our reviewer instructions. ~Kvng (talk) 21:06, 16 March 2026 (UTC)

Are you talking about pythoncoder's optional script or something else? ClaudineChionh (she/her · talk · email · global) 22:14, 16 March 2026 (UTC)

That's part of it and there's the LLM decline reason added to AFCH at some point and there's a pattern of rapid-fire declines using that reason by Pythoncoder and others. Reviewer instructions don't make any mention this, I don't know if authors are given fair warning, and yet it seems to be widespread practice. ~Kvng (talk) 22:22, 16 March 2026 (UTC)

• In line 6 of WP:YFA authors are indeed told not to use LLM. It's also in the "Don't" section. ChrysGalley (talk) 13:08, 17 March 2026 (UTC)

I've been reviewing AfC since around the middle of last year and I feel like the LLM decline was already present or may have been introduced at around the same time. It definitely predates NEWLLM. The WP:G15 speedy deletion criterion was added... I think after the LLM decline was introduced, but before NEWLLM. It probably makes sense to align the AfC quick-fail criteria with CSDs.

Reviewers are declining a lot of drafts for this reason because we see a lot of LLM drafts, but there is the danger that the more actual slop you see, the more predisposed you might be to see it everywhere. I haven't installed the one-click decline script because I don't want it to be so easy to make that call. ClaudineChionh (she/her · talk · email · global) 22:47, 16 March 2026 (UTC)

I can confirm that the LLM decline reason predates the introduction of G15, which in turn predates NEWLLM. The reason why I decline so many drafts for being LLM-generated is because there really are that many LLM-generated drafts. I don’t think it’s a coincidence that the last two AFC backlog drives have been just 6 months apart — and we’re already back up to about as many unreviewed drafts now as there were when the last drive started. As for the reviewer instructions, LLMs are fairly new in the grand scheme of things, so it’s only mildly surprising to me that that page hasn’t been updated yet. —pythoncoder (talk | contribs) 22:53, 16 March 2026 (UTC)

I’ve seen some reviewers outright reject LLM drafts, but I tend to go for a regular decline first because the submitter may not have been aware of the guideline. I recall that not too long ago I read through the instructions users are given before either creating or submitting their draft, and then suggested adding some guidance against using LLMs, but I don’t know what came of that. —pythoncoder (talk | contribs) 23:00, 16 March 2026 (UTC)

• I don't reject for AI, since the Reviewer Instructions states that is just for irredeemable scenarios. I decline a very high proportion based on AI authorship, because there is a lot of it presented here. But at least in theory the editor can simply re-write their draft themselves, though that is clearly wishful thinking in some cases. I don't also reject AI in borderline cases, or where just one section seems to be LLM. A key problem here is "Subject X was mentioned by A, B and C media outlets", so WP:AIATTR. That isn't summarising and arguably contrary to this project's purpose. This could be argued as a reject basis, but I've not done that.ChrysGalley (talk) 13:04, 17 March 2026 (UTC)

  Whether AI generated or not, WP:AIATTR is basically the advice I've been giving authors. I tell them that establishing notability in a NPOV manner with a short article is where we need to start especially if there is a COI in play. Have I been giving bad advice?

  Per the current reviewer instruction, we're not supposed to decline articles because they are stubby, don't have the right encyclopedic voice or have too many references. Improving these things is something that can get sorted out in mainspace and these flaws do not reflect particularly badly on Wikipedia and do not prevent readers from getting value from an article. ~Kvng (talk) 15:14, 17 March 2026 (UTC)

  @Kvng, I don't think you're giving bad advice per se. Really, you're giving the most important advice. But the community is so hostile to AI of any kind right now, I would be warning people that they really shouldn't be using it at all, right up front. It takes learning to understand how we implement NPOV. "For the love of god don't use AI until you already have a lot of experience with editing" is a simple and usually immediately comprehensible message.

  I agree that we should be adding some explicit guidelines regarding the AI decline on the reviewer instructions page. I would say that AI is a perfectly fine single-reason decline at the same level that npov is: that is, when the article needs so much work it simply should not be accepted and tagged. Someone who used AI to generate an article spent very little time on it. I don't want to ask reviewers to spend any more time than they have to on shovelling those out of the queue. -- asilvering (talk) 16:29, 17 March 2026 (UTC)

  The most frustrating thing with AI drafts is that even if it seems fine, the sources refer to the subject, etc, there's about a 75% chance of any given source not actually verifying the information it's attached to. That means AI drafts have to be gone over with a fine tooth comb and every single statement needs to be checked, which is a huge amount of work for the reviewer (and compounds the more times it goes through review). Human-generated drafts can be spot-checked, and if they've done it correctly in those places we can be pretty confident they know how to source. Plus of course humans frequently understand what needs to be fixed, and AIs...well, they do their best, and often offer some really fun new policies as justification just in case we like those better, so I guess at least there's amusement value. Meadowlark (talk) 08:28, 18 March 2026 (UTC)

  Actually, would it be worthwhile proposing something along the lines of this: if a draft is AI-generated, the first review gets declined with instructions that a human needs to rewrite it; if it's resubmitted and is still clearly AI-generated, it gets rejected. That might balance fairness to draft creators with fairness to reviewers. Meadowlark (talk) 08:34, 18 March 2026 (UTC)

  Apparently because I'm a robot, I personally am not good at distinguishing AI from human prose so I would be at a loss for how to rewrite so that it looks like it was written by a human. ~Kvng (talk) 15:04, 19 March 2026 (UTC)

I hope no one's actually accused you of being a robot! Learning to identify LLM text is just another skill that can be practised. See what other AfC reviewers have declined for this reason at Category:AfC submissions declined as a large language model output and try the AI or not quiz for practice. ClaudineChionh (she/her · talk · email · global) 23:08, 19 March 2026 (UTC)

No, I made the comment because other editors have made comments that LLM text is obvious and they are certain, based on the text alone, of it's providence. I'm skeptical of these claims because 1/ that is not my experience and, more importantly, 2/ AI is improving very quickly and I don't see any reason any obviousness won't soon be overcome and every reason to believe the certainty experienced by these human editors will likely linger. ~Kvng (talk) 01:31, 20 March 2026 (UTC)

@Kvng Would you say the same after reviewing this example? Your second point is well made, and so for example we don't see so many utm markers these days, but still LLM loves inventing our own categories. ChrysGalley (talk) 12:15, 20 March 2026 (UTC)

I got 50% on the AI or not quiz – no better than a coin flip. I don't trust myself to make these calls. ~Kvng (talk) 14:59, 20 March 2026 (UTC)

As I said, it's a skill that you can practise and get better over time. None of us are getting to 100% or even 70% accuracy without practice. ClaudineChionh (she/her · talk · email · global) 01:57, 21 March 2026 (UTC)

I've heard back from our alleged robot asserting that they are human. Does this change anything? Any advice on what advice I sould give this author? ~Kvng (talk) 16:52, 8 April 2026 (UTC)

Both the draft and the author's comments on your talk page, including the denial, absolutely reek of AI (to me). This happens often enough that we have a WP:NOLLMLIARS shortcut. The next step may be WP:AINB if you want to get a third (or more) opinion, or escalate to AN/I if you're feeling confident.

Something else to keep in mind going forward is that WP:NEWLLM has now been amended to cover all article content, not just new articles or drafts. ClaudineChionh (she/her · talk · email · global) 22:54, 8 April 2026 (UTC)

Honestly, nowadays I just assume everything from new editors is LLM-assisted unless proven otherwise. See Wikipedia_talk:New_pages_patrol/Reviewers#Am_I_crazy_or_are_the_majority_of_new_articles_LLM-assisted? for more info. I think it's getting harder to detect. Wikipedia:Signs of AI writing is basically just telling the LLM what not to do... lol. ~WikiOriginal-9~ (talk) 17:31, 8 April 2026 (UTC)

@WikiOriginal-9, so we should just stop accepting new articles? ~Kvng (talk) 21:17, 8 April 2026 (UTC)

No, I'm just advising caution. Thanks, ~WikiOriginal-9~ (talk) 21:46, 8 April 2026 (UTC)

Wow, tough crowd. I'm inclined to accept this draft and get a read on community sentiment in AfD. I assume at least one of you would be eager to nominate it. ~Kvng (talk) 02:32, 9 April 2026 (UTC)

Everyone's AI barometer is calibrated differently, so it's fair to accept a draft that you think should be accepted, as long as you're staying up-to-date with policies. But if you're referring specifically to Draft:Trinnov Audio, it hasn't actually been edited or resubmitted since @Pythoncoder declined it on 10 March. ClaudineChionh (she/her · talk · email · global) 02:58, 9 April 2026 (UTC)

I mean, AGF it's human-written, does it pass NCORP? Assuming we're talking about Trinnov, everything I'm seeing is either non-independent, not about the company (the product reviews), or WP:CORPTRIV. But I'm sick and a bit whirly so I may have missed something. Meadowlark (talk) 06:23, 9 April 2026 (UTC)

Draft:Trinnov Audio. I believe it is WP:UNLIKELY to be deleted for notability concerns. We have identified WP:THREE independent sources. See User talk:Kvng#Inquiry about AfC drafting support for tech company article for details. I'd encourage any discussion specific to the contents of this draft occur on the draft's talk page. ~Kvng (talk) 14:40, 9 April 2026 (UTC)

Hey all, I have been slogging through the AfC queue today and I don't feel confident in noticing the tell-tale AI signs. I applaud the numerous users on here who can call a spade a spade from miles away. I have been able to (at the very least) check links to make sure they aren't hallucinated/404, but I still feel behind. I know that there are several sites on the internet that can tell you what percentage something is written by an LLM. Is this something that we could incorporate into a list like User:RichBot/copyvios does for Copyright Violations using Earwig's CV detector? Thanks for any thoughts, and keep fighting the good fight against slop! Bkissin (talk) 22:50, 28 March 2026 (UTC)

I think the two key problems here are that the best detection software are paid-for products (Winston, GPTZero, Pangram, in my view), so there's a finance issue, someone has to negotiate and then pick up the bill. The free software such as Zerogpt is nothing like as good, false positives and false negatives. The second issue is that with Copyvio, there is only one way out, the editor has to summarise rather than lift the source (hopefully not using LLM....); whereas with LLM usage there are two ways out of the LLM rabbit-hole: a human re-writes the entire article from scratch, given that the recently revised WP:NEWLLM has no wriggle room for 99% of drafts that are not direct language translations. Alternatively the LLM just gets smarter and directly defeats the checks that we deploy, particularly if like Earwig it is a public resource. ChrysGalley (talk) 10:15, 29 March 2026 (UTC)

There used to be a wikipedia.gptzero.me, but no longer. I've asked WMF staff to plug this hole, to no avail. One thing you can look at - which still won't catch all of it because it's not enabled for everyone and it only works for people using Visual Editor - is the tags of the edits, to see if "paste check" shows up. If it does, you know that the author has copy-pasted something into the draft. Of course, that can be done innocently, for example if you've written the draft offline or have copy-pasted quotes from sources. But it's another "objective" tell. Eventually, you'll just get good at spotting the AI signs yourself.

It only really matters if something is AI if it fails our content policies. So if an article's links aren't hallucinated and they verify the text, you don't need to worry about it. -- asilvering (talk) 11:44, 29 March 2026 (UTC)

@Bkissin, I know I'm very late to respond, but the most consistent (and dangerous) AI tell is citations to plausible sources that do exist but do not actually verify the claim. A draft I reviewed recently shows this well: Draft:Valentina Peleggi. In the 'Guest conducting' section, the draft claims The Guardian praised her performance, noting she conducted with "marvelous flair and precision.". The Guardian article does exist and is about the performance, but the quoted text is not in the article. Down in Discography a similar problem occurs: the draft claims that Gramophone reviewed the album favorably, noting Peleggi's "astute direction" of the choir. Gramophone did review the album, but the quoted text is from a different review that's also been collated on the same page. There may be more but at that point I declined the draft. Meadowlark (talk) 03:46, 16 April 2026 (UTC)

I accepted a draft that was in a category for drafts. This resulted in the article being in a category for drafts, and the draft category including an article. I removed the article from the category, and tagged the article for {{improve categories}}, but have a question about automated or semi-automated cleanup. If I understand correctly, the move from the draft namespace to the article namespace created what is known as a polluted category. Are these polluted categories either automatically cleanup by a bot, or listed in a report so that they can be cleaned up by gnomes? Robert McClenon (talk) 16:07, 31 March 2026 (UTC)

There are some database reports on polluted categories, yes. I'm not sure there are any bots specifically for draft-cats-in-articles, though. Primefac (talk) 23:46, 31 March 2026 (UTC)

hey guys, I'm just noticing that AfC submissions is adding up every time, it made me question if this is an all time high? if not when was it and how did we manage to reduce it. Cmajor^ftw 14:14, 1 April 2026 (UTC)

In the Nov-23 drive we started with a huge backlog, did over 12K reviews, and brought it right down to zero. -- DoubleGrazing (talk) 14:23, 1 April 2026 (UTC)

High, but not an all time high, it has been over 5000. However, it is still depressing so see it climb so high so quickly. KylieTastic (talk) 15:08, 1 April 2026 (UTC)

Isn't February particularly active for article creation and November not as much? That might explain the plateau on one side and the huge rise on the other. ✶Quxyz✶ (talk) 12:45, 4 April 2026 (UTC)

I wonder about the concept of auto-declines for those with no content of no references. It's not a large population, but some of the load would go away.

In the emerging days of AI we could also look at 'silliness' before it gets to a reviewer 🇵🇸‍🇺🇦 Fiddle^Timtrent Faddle^Talk to me 🇺🇦‍🇵🇸 15:17, 1 April 2026 (UTC)

There's so much AI it's ridiculous. I think lot's of stuff goes unnoticed to. It's getting harder to detect. ~WikiOriginal-9~ (talk) 19:12, 1 April 2026 (UTC)

seems like a good idea the auto decline Cmajor^ftw 19:24, 1 April 2026 (UTC)

Nah... it has yet to reach the 6 months limit. – robertsky (talk) 15:26, 1 April 2026 (UTC)

@robertsky don't you dare jynx it. -- asilvering (talk) 18:59, 1 April 2026 (UTC)

@Asilvering its a slo... – robertsky (talk) 19:45, 1 April 2026 (UTC)

The prior thread made me wonder: when we experience precipitous increases in the backlog, is it the result of decreased reviewing, increased submissions, or both? We already have a running chart for the backlog, is there any way for us to add another graph with two lines: reviews, and new submissions. We of course already track the new submissions number. But how hard would it be to track the reviews? I'm thinking we don't necessarily have to differentiate between kind of review, just at the start of the day, did Draft:Foo have an unreviewed submission, and did it not at the end of the day? CaptainEek ^{Edits Ho Cap'n!}⚓ 19:48, 1 April 2026 (UTC)

Or I guess the simpler mathematical approach is just (amount backlog changed from yesterday) - (submissions) = |reviews| CaptainEek ^{Edits Ho Cap'n!}⚓ 20:03, 1 April 2026 (UTC)

It's difficult to get exact numbers here because of deletions, resubmissions, and drafts that are reviewed on day 0 (which is most drafts). -- asilvering (talk) 20:23, 1 April 2026 (UTC)

I used to keep some fairly in-depth stats about this sort of thing, and the short answer is basically "on average, we get more submissions than we review". I know this seems somewhat obvious (since new minus reviewed is clearly positive) but that is kind of The Answer. As a note, I do say "on average", because while some months we have more reviews than submissions, over a long enough time period those small gains are outweighed by others months with more submissions. Primefac (talk) 20:31, 1 April 2026 (UTC)

Just to flag that @DoubleGrazing: and I have both had AFCH not properly pasting declines on to the submit editors' Talk pages. Inconsistent since my first one worked, and DG's did not on the same draft. Approx 1 in 5 occurrences in my case. It then means reviewers may want to do a manual entry instead. ChrysGalley (talk) 11:10, 2 April 2026 (UTC)

How quickly do you move away from the page after you have declined? DGG used to have this problem in the past as he would often decline and then (almost) immediately close the window before things actually went through. Does the script say that the user talk message has been left? Primefac (talk) 11:34, 2 April 2026 (UTC)

Only happened to me once (that I know of), and no I didn't move away from the page quickly. The script said something to the effect of failed to notify user, API db error summat summat. -- DoubleGrazing (talk) 11:39, 2 April 2026 (UTC)

Same for me and I think I've had 3 occurrences out of 18 reviews this UK morning. I don't think it's about speed at my end, since the error "api DB" appears while the other lines are getting populated, so the error is shown before I see the line about my AFC log file getting updated. Has a latency feel to it. ChrysGalley (talk) 11:47, 2 April 2026 (UTC)

Just checked and your AfC log has grown quite long; may be worth setting up archiving to see if that might fix things? nil ^nz 11:51, 2 April 2026 (UTC)

Is it just within the 90 minutes or so? We're in the middle of a minor outage right now, judging from the comments left on phab:T422130. –Novem Linguae (talk) 11:55, 2 April 2026 (UTC)

My decline is timestamped 12:01 (UTC+1, presumably). -- DoubleGrazing (talk) 12:41, 2 April 2026 (UTC)

Ironically enough, as I was posting that comment, I got [195e7548-6cce-485c-9cc8-96f07eb363ef] Caught exception of type Wikimedia\Rdbms\DBConnectionError -- DoubleGrazing (talk) 12:43, 2 April 2026 (UTC)

Gonna make me math, are ya? Do you think that was within about 2.5 hours of now, or earlier? :) –Novem Linguae (talk) 13:03, 2 April 2026 (UTC)

I don't know! I 'spose. (Maybe.)

When we (UK) switch to daylight saving time, some timestamps are in UTC, some are in UTC+1, and I've no idea why that is, or whether I can do something about it. The ones on talk pages like this very helpfully say 'UTC', but the contribs list and rev history timestamps say nothing.

Anyway, I declined a draft maybe 10 mins ago, and now realise the decline didn't get posted on the user's talk, so this may still be ongoing? -- DoubleGrazing (talk) 13:33, 2 April 2026 (UTC)

Yeap. Still ongoing. I had to press "reply" to my last message 5 times to get it to post. –Novem Linguae (talk) 13:39, 2 April 2026 (UTC)

 You are invited to join the discussion at WP:VPR § A new system of permissions for article creation. This might be of interest for AfC reviewers. Athanelar (talk) 15:01, 4 April 2026 (UTC)

Moved from Wikipedia talk:WikiProject Articles for creation/Reviewing instructions/Copyright cleanup instructions

Point (iv) here doesn't seem correct. If an editor changes the template param from cv to cv-cleaned then the message on the page then displays

"This submission has now been cleaned of the above-noted copyright violation and its history redacted by an administrator to remove the infringement. If re-submitted (and subsequent additions do not reintroduce copyright problems), the content may be assessed on other grounds."

The second point about the history is obviously not the case until the revdel has been completed.

I submit that point (iv) should be removed entirely, as the admin performing the revdel should change that after completing the process. The copyvio tag should be left as cv. Mfield (Oi!) 00:13, 5 April 2026 (UTC) Link to the cv cleanup added for clarity, no content change. Primefac (talk) 16:59, 5 April 2026 (UTC)

Or alternately it's the AFC template that needs correcting so it doesn't display that, if editors are being encouraged to flag the copyvio template as cv-cleaned. Mfield (Oi!) 00:56, 5 April 2026 (UTC)

Point iv cannot be taken in isolation. The preceding steps involved a) removing the copyvio and b) tagging the page for {{revdel}}. An admin will remove the revdel request when the content has been hidden, but from a non-administrative perspective, the draft has been cleaned of copyright violations. Primefac (talk) 17:01, 5 April 2026 (UTC)

OK point noted about administrative vs not. I am only looking at it through an admin lens. Maybe it could be reworded to say

"This submission has now been cleaned of the above-noted copyright violation and its history will be redacted by an administrator to remove the infringement. If re-submitted (and subsequent additions do not reintroduce copyright problems), the content may be assessed on other grounds."

Mfield (Oi!) 18:36, 5 April 2026 (UTC)

Sounds sensible to me. I always found that decline very confusing as a reviewer. -- asilvering (talk) 03:55, 6 April 2026 (UTC)

Ah, okay, I see the issue. cv says "clean this up now!" whereas cv-cleaned implies that all stages of cleanup have occurred (including revdel which may still be pending). I think the better solution would be to word along the lines of "cv has been removed, but if it isn't RD'd yet please switch it back to a cv decline" maybe? That way the administrator only has to do the revdel without worrying about messing about with the template, but interested editors can still do a quick check if they so desire. Primefac (talk) 23:32, 6 April 2026 (UTC)

Yep, I don't really have any strong opinion on which aspect of it should be changed, except that the reviewer instructions and the displayed message end up aligning and reflecting the actual status. My anecdoatal experience of dealing with RD1 requests (both AFC and others) is that most reporters set the tag only to cv. The situation that brought this up for me the other day was that I responded to a couple of reports where the same reviewer has set them to cv-cleaned already. This appeared unusual to me so I asked them about it and they pointed me to the instructions. That's the reason I figured that the instructions needed updating tbh, as they didn't seem to match the spirit of the tags or the way most editors interpret them at least. Mfield (Oi!) 01:00, 7 April 2026 (UTC)

I had previously assumed the administrator doing the revdel was supposed to change it from cv to cv-cleaned to guarantee it had been done - is there an edit filter or something that prevents removal of the {{cv-revdel}} template? ScalarFactor (talk) 03:05, 7 April 2026 (UTC)

That is what I always assumed also yes and that is what I do in other circumstances when I am needing to leave the copyvio template for some other reason. But when i complete an RD1, normally via script, hitting complete on the process removes the copyvio template anyway.

The primary concern here though is that the instructions tell reviewers to set the template to cv-cleaned when before the revdel is actually completed. So when the admin responding, or anyone for that matter, looks at the article, the AfC decline box shows a message saying that the article history has been cleaned when it hasn't. So really it's about changing that instruction so that the copyvio template gets put on as just cv. Which seems to be a common practice already outside of editors who are strictly following the instructions, as maybe a lot of those people realized that incorrect message then appears. Mfield (Oi!) 03:25, 7 April 2026 (UTC)

To answer the other part of the question, there's no edit filter preventing a {{cv-revdel}} template from being removed, but if the person removing it isn't an admin or NPR it'll be tagged by filter 856. That filter gets patrolled semi-regularly so it'll likely be caught eventually if someone does remove the template. Obviously it's ideal for it to be fixed immediately though, so personally I always temporarily watchlist pages that I've tagged for RD1. MCE89 (talk) 03:31, 7 April 2026 (UTC)

Is there an option to undo a waiting submission? As a mentor, I sometimes advise mentees that their drafts are hopeless (not in those words) and rather than wait another month or two for the inevitable decline, they should withdraw it and work on the draft now rather than months down the road. Withdrawal would also save an Afc reviewer from wasting their time. This was discussed once in 2009 and opinions then were mostly that it was too rare a situation to bother about.

It isn't that rare now, imho, as I come across this situation frequently. Any interest in having another look at this? Alternatively, if my mentee agrees, I am inclined to implement the withdrawal by simply removing the timestamped {{AFC submission}} template at the top of the page (the one that has the hidden comment, <!-- Do not remove this line! --> ) and replacing it with {{AfC submission/draft}}. Otoh, if the timestamped submission causes some external action such as adding a timestamped control field to an entry in a database or project page that would then be orphaned by removal of the template with the matching timestamp, then I won't, but otherwise, I cannot see any reason not to remove the line and save everybody from wasting their time. Mathglot (talk) 21:13, 10 April 2026 (UTC)

I don't believe there's any issues removing AfC templates like that. We generally tell editors not to remove previous declines, but as an optional process for autoconfirmed editors they're free to remove the templates and just move it to mainspace themselves, for example. ScalarFactor (talk) 00:37, 11 April 2026 (UTC)

AFAIK, reverting the submission (if still possible) will just withdraw it from the pool as if it was never there. Or you could manually delete the AfC templates altogether. I don't think either would cause any problems, and if it's clear (say, from an edit summary) that it was done for a reason, then I doubt anyone here would try to undo such an edit or otherwise take issue with it.

FWIW, I for one very much welcome such input, as well as commending you for making an effort to work with novice users in this way. Submitting a draft prematurely helps no one and only causes frustration and extra work all around. Plus, with guidance from an experienced user such as you, the new editor will surely go through a much smoother learning curve, which can only be a good thing. -- DoubleGrazing (talk) 05:37, 11 April 2026 (UTC)

Three options, all of which are valid:

• Undo the edit that submitted the draft (could be blocked by intermediate edits)
• Add t to the first parameter of the {{AfC submission}} template (e.g. {{AfC submission|t|...}})
• Replace the {{AfC submission}} template entirely with {{AfC submission/draft}}

At the moment there is no script-assisted way to perform any of these actions, so it will have to be done manually. Personally my preference would be for the third option since it makes for a cleaner resubmit if/when that comes. Primefac (talk) 11:32, 14 April 2026 (UTC)

Thank you, both. Bullet #3 (replace with {{AfC submission/draft}} ) was my intended action (I have it in my tips file as the end result of {{subst:AfC draft}} ), so I will likely go with that one. I'll look into adding something like this to the doc somewhere. I think it may more properly belong on the WP:Mentorship page than on an Afc doc page as it is out-of-process for Afc, but appreciate any suggestions on where it should live. I'm thinking a brief mention or link at Help:Drafts (perhaps at § Moving drafts to mainspace ) and WP:User pages (perhaps § Userspace and mainspace ). Mathglot (talk) 17:52, 14 April 2026 (UTC)

Since it's a bit more procedural than technical it might be better to put it on the mentorship or help pages, but if it's a source of confusion it definitely should go somewhere! Primefac (talk) 11:08, 15 April 2026 (UTC)

I recently noticed it is possible for temp accounts to create draft articles. Given that article creation by IP addresses was banned in the wake of the Wikipedia Seigenthaler biography incident, as well as the possibility of temp accounts creating hoax draft articles, I'm not sure why they can do this. JHD0919 (talk) 20:10, 15 April 2026 (UTC)

Allowing IPs, TAs, and non-autoconfirmed users to create articles is more or less the entire purpose of the Draft namespace. It's also worth noting that Draft is NOINDEXed and always has been. —Jéské Couriano v^_^v ^MUSHROOM 20:23, 15 April 2026 (UTC)

Judging from the history essay, I'll give you that's the purpose of the Draft namespace. However, I fail to see how the drafts being NOINDEXed is relevant to my concerns about temp accounts being able to make them. JHD0919 (talk) 21:53, 15 April 2026 (UTC)

What is your concern about TAs creating drafts? (which, as noted, is pretty much the main purpose of draft space) 331dot (talk) 21:56, 15 April 2026 (UTC)

the possibility of temp accounts creating hoax draft articles. I also have concerns about temp accounts creating articles solely based off of misinterpretations of the referenced sources, as is the case with Draft:Bee Movie 2 (which I recently took to MfD for that very reason.) JHD0919 (talk) 22:06, 15 April 2026 (UTC)

I mean registered accounts do the exact same thing, and not everything TAs make is a hoax so stopping TAs from creating drafts isn't stopping hoax drafts and is just eliminating the potential for the good articles they do create. Sophisticatedevening^(talk) 22:10, 15 April 2026 (UTC)

Fair enough. JHD0919 (talk) 22:12, 15 April 2026 (UTC)

The Wikipedia Seigenthaler biography incident was 20 years ago so is not exactly the most contemporary BLP policy issue to solve anymore. But anyway, part of why draftspace was created was to allow folks that got locked out of creating mainspace articles during WP:ACTRIAL to have a place to create articles. So I think draftspace is working as designed.

How much to restrict newer editors from submitting poor articles is frequently discussed at places such as the village pump. It's a hot button issue that dozens of experienced editors have strong opinions on, but the current equilibrium has remained stable for many years. –Novem Linguae (talk) 21:59, 15 April 2026 (UTC)

Hi, I am reviewer at AFC and created a draft Draft:Atal Canteen scheme, can I review it myself :) or leave it there until reviewed. ✓ortexPhantom (talk) 10:54, 16 April 2026 (UTC)

Unless you have a COI, AfC is completely optional for experienced editors – you're more than welcome to move it to mainspace yourself if you believe it's ready :) nil ^nz 11:43, 16 April 2026 (UTC)

Ok, I understand, I will move to mainspace. ✓ortexPhantom (talk) 11:56, 16 April 2026 (UTC)

Feel like something that'd be helpful for reviewers is if submitters were required to write a brief explanation of how their article meets notability requirements AND if there was a checklist they'd have to manually check the boxes that certifies they understand the most important aspects of notability/NPOV/COI etc. The checklist would provide links to each policy.

These should be written/presented as succinct as possible, so that it's basically only a minute of labor on part of submitter. Shouldn't be cumbersome. The explanation would appear to reviewer; this would help direct reviewer's attention to like e.g. the three pieces of SIGCOV that prove notability or like the proof that some academic meets NACADEMIC.

A lot of the time it's annoying for reviewers to dig through valid uses of primary sources or press releases etc to find the 3 pieces of SIGCOV that push an article over the line. If the submitter identified those for us it'd help somewhat. Also the presentation of a checklist and forcing people to write out/think about what notability is I think would help. grapesurgeon (talk) 02:01, 20 April 2026 (UTC)

I think notability is probably too complicated for this to work. Guidelines such as WP:GNG are too vague (compared to how it actually works). The various WP:SNGs are too specific, with very uncommon SNGs mixed in with common ones, adding a lot of noise. These two things make it hard to teach beginners how notability works, and hard to trust beginners to be able to accurately judge notability. Therefore, even if you make a checklist that has the item "My draft contains 3 sources with significant coverage", I think beginners would misjudge their sources more often than not, thus not really improving the situation. –Novem Linguae (talk) 02:42, 20 April 2026 (UTC)

I hold similar reservations about this plan (which in fairness, sounds like a good idea) based on my own past experience with WP:THREE. When asked for the best refs on IRC, helpees will invariably give me a half-dozen garbage sources. While it might not be difficult for the draft submitter to write up, reading an extra paragraph or two of pseudo-abstract and checking three potentially-not-useful references will likely just add to the review time. I would rather skim the draft and the sources and make a judgement based on that.

Conversely, this could potentially harm people who make a poor argument for what would otherwise be a well-written draft, because it might unduly influence the reviewer (especially if they assume that the three best really are the three best, even if they're not). Primefac (talk) 18:19, 20 April 2026 (UTC)

I'm not sure if this is the best place to report this problem, but the "Clean submission" link in the helper script submits unsubmitted drafts. Specifically, if the draft contains the template {{AfC submission/draft}} and a reviewer clicks "Clean submission", the template is changed into {{AFC submission||ts=}} <!-- Do not remove this line! -->. The new submission lacks both a submitter and a timestamp which creates other problems. I have no idea if this is a new issue or if it has existed for some time, but in any case it's not how the script is supposed to behave! And the reviewer who has cleaned up the page doesn't see the new submission unless they reload the draft page, so it's extremely easy to miss. Here is one example; I also experimented with a test page in draftspace but I have tagged that for deletion. --bonadea contributions talk 09:18, 21 April 2026 (UTC)

Likely a regression, this specific bug was fixed at the New Year but I'm not sure what would have caused it. Primefac (talk) 11:14, 21 April 2026 (UTC)

I came across Draft:Bachelor of Computer Applications, searched the web for the qualification name, and found the draft had been copied to Bachelor of Computer Applications, which I've nominated for PROD. I've now dug a bit more and realised that this brand-new editor with seven edits was able to "publish" their draft because there was an existing redirect with that name, which they could simply overwrite. I don't think I've seen this happen before so maybe it's not a big deal, but should there perhaps be a filter to catch non-confirmed users effectively bypassing the autoconfirmed requirement this way (even if it's by accident)? ClaudineChionh (she/her · talk · email · global) 09:01, 24 April 2026 (UTC)

(And now that I've typed tnis out, I think maybe I could have just reverted to the redirect in this case, instead of PROD?) ClaudineChionh (she/her · talk · email · global) 09:03, 24 April 2026 (UTC)

It does happen occasionally. Now that you've put on a PROD you might as well leave it, but I do think that reverting it back to a redirect probably would have sufficed. Primefac (talk) 10:20, 24 April 2026 (UTC)

I'll try to watch out for that now that I've seen it once. ClaudineChionh (she/her · talk · email · global) 14:27, 24 April 2026 (UTC)

Redirects flipped to articles get automatically unreviewed, so will enter the NPP queue and hopefully get caught that way. –Novem Linguae (talk) 15:37, 24 April 2026 (UTC)

I've just been declining them with the LLM reason + relevant notability reason, but I've seen other reviewers CSD them as well/instead. Should I also be CSDing them, or is this more of a personal choice situation? This is assuming they're clearly AI with the relevant problems ie nonsense refs, commentary to the user, and so on. Meadowlark (talk) 03:26, 25 April 2026 (UTC)

If they can be cleaned up, decline them. If they are irredeemable, delete them. Primefac (talk) 09:00, 25 April 2026 (UTC)

Also worth noting that the criteria for WP:G15 is pretty narrow, and not every LLM draft will meet those criteria. From personal experience, most LLM drafts have real references (so aren't G15-able), but those refs don't actually support the content it's cited for. nil ^nz 09:08, 25 April 2026 (UTC)

I've also given up on using G15 (and other speedies for that matter), a proportion of them are not upheld, it's an inefficient use of resources, and sends mixed messages to the LLM editor. There is some advantage in keeping the dodgy draft, to see what happens next. Very occasionally the editor or someone else will in fact rewrite the draft, and you get to see the fingerprints, whereas deleted speedies become guess work for non administrators. If the project is being damaged then I will go the extra mile but usually LLM is just lazy editorship. ChrysGalley (talk) 12:15, 25 April 2026 (UTC)

I find the use of these words confusing. Normally, these two words have basically the same meaning, but they have particular and distinct meanings in AFC. Could we reword these? Maybe "decline" and "decline no resubmit", alternatively "reject" and "reject no resubmit". grapesurgeon (talk) 14:24, 25 April 2026 (UTC)

You may be interested in this RFC from September (and it's accompanying Village Pump discussion). My personal feeling is that, regardless of terminology, for new editors a big red box that tells them their draft wasn't accepted = "rejected". nil ^nz 14:49, 25 April 2026 (UTC)

Whenever I visit my own userpage I get the following notice:

AFCH error: user not listed AFCH could not be loaded because "Coolgurl5555" is not listed on Wikipedia:WikiProject Articles for creation/Participants. You can request access to the AfC helper script there. If you wish to disable the helper script, click here. If you have any questions or concerns, please get in touch!

Of course, I clicked on the “please get in touch” link. - coolgurl5555 ✈︎ (she/her) ✈︎ 15:32, 25 April 2026 (UTC)

That is because you are not on the list of reviewers. I do note you have a pending application, which if another admin does not get to first I will look at tomorrow. Primefac (talk) 15:36, 25 April 2026 (UTC)

Yep I realized that immediately after posting this. I’m fucking stupid. Sorry! - coolgurl5555 ✈︎ (she/her) ✈︎ 15:37, 25 April 2026 (UTC)

pat pat, we all make mistakes! No worries :) GGOTCC 15:41, 27 April 2026 (UTC)

Hello,

I submitted my draft (Draft:Andzejs_Grauds) several months ago and would greatly appreciate if someone could review it.

The article includes multiple independent sources and references.

Thank you very much for your time and support.

Best regards, Andzejs Grauds Bunguskola (talk) 11:00, 28 April 2026 (UTC)

Bunguskola This page is for discussion about the AFC process itself, and is not for requesting draft reviews. Please use the AFC help desk to ask questions related to a draft. The time estimate on the draft is just that, an estimate, and the process may take longer depending on the willingness of volunteers to do the work. Asking for a review will not speed the process, please continue to be patient. 331dot (talk) 14:51, 28 April 2026 (UTC)

The script should not insert colons into category links if they are in a block of {{draft categories}} because that defeats the purpose of that template. -- Michael Bednarek (talk) 02:19, 29 April 2026 (UTC)

Could you please give an example? I feel like this issue was dealt with a while ago but we've had a few weird regressions lately. Primefac (talk) 09:06, 29 April 2026 (UTC)

The ticket is still open, so it probably hasn't been fixed yet. –Novem Linguae (talk) 11:36, 29 April 2026 (UTC)

I've seen it occasionally after I placed {{draft categories}} in articles. This prompted me raising it here. -- Michael Bednarek (talk) 12:34, 29 April 2026 (UTC)

Fixed. –Novem Linguae (talk) 20:08, 1 May 2026 (UTC)

Hi, I just declined Draft:Chase Griffin (author), but the helper script tried sending the decline message to nonexistent user "YourUsernameHere" which is also on the blacklist, but the real submitter was User:~2026-25990-18, I'm not sure what happened, looking at the AfC submission template it appears the submitter name was correct and they used the submit wizard. Does anyone know what may have happened? 🌀Hurricane Wind and Fire, why did you decline my draft? (talk) (contribs)🔥 03:12, 29 April 2026 (UTC)

The TA manually input "YourUsernameHere" with this diff and likely submitted it as that user. • Quinn (talk • it/its) 03:21, 29 April 2026 (UTC)

Thanks, I'll try to catch that in the future. 🌀Hurricane Wind and Fire, why did you decline my draft? (talk) (contribs)🔥 03:22, 29 April 2026 (UTC)

I've at least adjusted that part of the submission template so that shouldn't (theoretically) happen again. I'm inclined to believe whatever LLM they used to write the draft is what made that change. • Quinn (talk • it/its) 03:29, 29 April 2026 (UTC)