Wikipedia talk:WikiProject Spam
From Wikipedia, the free encyclopedia
 |
When reporting spam, please use the appropriate template(s):
|
As a courtesy, please consider informing other editors if their actions are being discussed.
Also, please include links ("diffs") to sample spam edits. | |
 | WikiProject Spam was featured in a WikiProject Report in the Signpost on 18 July 2011. |
| Indicators |
|---|
| Reports completed: |
 Done |
 No action |
 Stale |
| Defer discussion: |
 Defer to XLinkBot |
| Defer to Local blacklist |
| Defer to Global blacklist |
| Defer to edit filter |
| Information: |
 Additional information needed |
 Note: |
Suspicious new articles
List of reports here. Remove transclusions when all finished to keep page readable.
Suspicious new articles (2026-03-07)
2026-03-07: Break 0
2026-03-07: Break 1
2026-03-07: Break 2
2026-03-07: Break 3
2026-03-07: Break 4
2026-03-07: Break 5
2026-03-07: Break 6
2026-03-07: Break 7
2026-03-07: Break 8
2026-03-07: Break 9
2026-03-07: Break 10
Suspicious new articles (2026-02-14)
2026-02-14: Break 0
2026-02-14: Break 1
2026-02-14: Break 2
2026-02-14: Break 3
2026-02-14: Break 4
2026-02-14: Break 5
2026-02-14: Break 6
2026-02-14: Break 7
2026-02-14: Break 8
User:HuetteyWaa persistent link spamming
- HuetteyWaa (talk · contribs · deleted contribs · nuke contribs · logs · filter log · block user · block log) – Persistent link spamming of "WahlenWeltWeit" across multiple election articles. Single-purpose account, site lacks an imprint and reliability. PolitikNerd42 (talk) 09:43, 15 March 2026 (UTC)
cipscourse.com
- cipscourse.com: Linksearch en (insource) - meta - de - fr - simple - wikt:en - wikt:fr • Spamcheck • blacklist hits • MER-C X-wiki • gs • Reports: Links on en - COIBot - COIBot-Local • Discussions: tracked - advanced - RSN • COIBot-Link, Local, & XWiki Reports - Wikipedia: en - fr - de • Google: search • meta • Domain: domaintools • AboutUs.com
- Flomy-786 (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)
- Almas-7866 (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)
Not ready for us to do anything about this link to a training course in Dubai yet, but since the user resorted to socking to continue to add it to Chartered Institute of Procurement & Supply it seems worth noting it here for the archives in case it gets worse in future. • a frantic turtle 🐢 13:28, 16 March 2026 (UTC)
- Noting for the record that Flomy-786 (talk · contribs) returned today and was blocked for spamming the link again. • a frantic turtle 🐢 12:35, 18 March 2026 (UTC)
- @A Frantic Turtle Apologies for the slow reply. The two accounts are
Confirmed/ Technically indistinguishable. I've blocked both of them and the IP they shared. I'm not inclined to do anything else unless they come back. HJ Mitchell | Penny for your thoughts? 20:23, 20 March 2026 (UTC)
- @A Frantic Turtle Apologies for the slow reply. The two accounts are
redeagle.tech
- redeagle.tech: Linksearch en (insource) - meta - de - fr - simple - wikt:en - wikt:fr • Spamcheck • blacklist hits • MER-C X-wiki • gs • Reports: Links on en - COIBot - COIBot-Local • Discussions: tracked - advanced - RSN • COIBot-Link, Local, & XWiki Reports - Wikipedia: en - fr - de • Google: search • meta • Domain: domaintools • AboutUs.com
- mm2.bet: Linksearch en (insource) - meta - de - fr - simple - wikt:en - wikt:fr • Spamcheck • blacklist hits • MER-C X-wiki • gs • Reports: Links on en - COIBot - COIBot-Local • Discussions: tracked - advanced - RSN • COIBot-Link, Local, & XWiki Reports - Wikipedia: en - fr - de • Google: search • meta • Domain: domaintools • AboutUs.com
- Farazvilla (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)
Not sure if it's worth blacklisting these yet, but some blocks and maybe a CU would be helpful. It's my first time on WPSPAM so let me know if there's anything else I should do :) lp0 on fire () 10:19, 21 March 2026 (UTC)
- Actually, better idea: can we just blacklist everything ending in .bet? It's probably all online gambling spam and the few legitimate domains can be whitelisted. lp0 on fire () 11:05, 21 March 2026 (UTC)
- @Lp0 on fire I ran a quick check and found
- Farazvilla (talk+ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI)
- Dashawn59 (talk+ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI)
- Ikonopelski (talk+ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI)
- Amor7539 (talk+ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI)
- Budals49645 (talk+ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI)
- Casting49854 (talk+ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI)
- Huelbert13 (talk+ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI)
- Bangkok64 (talk+ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI)
- Looking a bit further, we also have Seomaster234 and CPMSEO, both now blocked, and a few temporary accounts. HJ Mitchell | Penny for your thoughts? 16:25, 21 March 2026 (UTC)
- @Lp0 on fire I've added mm2.bet to WP:BED and requested global blacklisting for redeagle.tech because that one has also been spammed on the Spanish Wikipedia. As far as I can tell, all registered involved are blocked, as are the underlying IPs of the temporary accounts. HJ Mitchell | Penny for your thoughts? 19:20, 21 March 2026 (UTC)
- @HJ Mitchell: Thanks. Thoughts on blacklisting all of .bet, like we do for .xyz? lp0 on fire () 19:36, 21 March 2026 (UTC)
- Personally I'm all for it based on anecdotal evidence but we probably need some actual data before blacklisting a whole TLD. HJ Mitchell | Penny for your thoughts? 20:17, 21 March 2026 (UTC)
- @HJ Mitchell: Thanks. Thoughts on blacklisting all of .bet, like we do for .xyz? lp0 on fire () 19:36, 21 March 2026 (UTC)
- @Lp0 on fire I've added mm2.bet to WP:BED and requested global blacklisting for redeagle.tech because that one has also been spammed on the Spanish Wikipedia. As far as I can tell, all registered involved are blocked, as are the underlying IPs of the temporary accounts. HJ Mitchell | Penny for your thoughts? 19:20, 21 March 2026 (UTC)
- @Lp0 on fire I ran a quick check and found
levantleather.com
- Bilary Hina (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)
- levantleather.com: Linksearch en (insource) - meta - de - fr - simple - wikt:en - wikt:fr • Spamcheck • blacklist hits • MER-C X-wiki • gs • Reports: Links on en - COIBot - COIBot-Local • Discussions: tracked - advanced - RSN • COIBot-Link, Local, & XWiki Reports - Wikipedia: en - fr - de • Google: search • meta • Domain: domaintools • AboutUs.com
The only other addition of this link is by a locked spambot on simplewiki. Could we run a CU, since the other account also added another domain, and there are are likely to be more? lp0 on fire () 10:44, 23 March 2026 (UTC)
- @Lp0 on fire you were right. CU is enwiki-specific so I can't see any accounts that only exist on Simple or elsewhere but the only other account on the IP was Fisgy Dora (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), another spammer who is technically identical. I wouldn't be surprised if it goes deeper but they're using a proxy (which I've blocked, along with both accounts). HJ Mitchell | Penny for your thoughts? 18:21, 23 March 2026 (UTC)
- Thanks. I've reported the IP for a gblock since it's a proxy; hopefully that will pick up a few more. lp0 on fire () 19:04, 23 March 2026 (UTC)
massarservic.net
- massarservic.net: Linksearch en (insource) - meta - de - fr - simple - wikt:en - wikt:fr • Spamcheck • blacklist hits • MER-C X-wiki • gs • Reports: Links on en - COIBot - COIBot-Local • Discussions: tracked - advanced - RSN • COIBot-Link, Local, & XWiki Reports - Wikipedia: en - fr - de • Google: search • meta • Domain: domaintools • AboutUs.com
Could we run a CU on this as well? There's definitely at least two accounts. I'm not entirely confident that there will be more spamming other domains, but it's plausible. lp0 on fire () 10:50, 23 March 2026 (UTC)
- @Lp0 on fire guess what? Jinefar Fees from this group is confirmed to Fisgy Dora from above, but on a different range! Ktyra Ega is confirmed to them and Xamlu.sn is
Possilikely (a mix between possible and likely) the same. The other account, Utawr Gyta, is on a different range in a different country. If we see the link again, it would probably be ripe for local or possibly global blacklisting. HJ Mitchell | Penny for your thoughts? 20:36, 23 March 2026 (UTC)
- That's genuinely a little surprising. Since CUs get GTAIV, could you find anything here on the same IP as the urwiki addition in January? lp0 on fire () 20:41, 23 March 2026 (UTC)
- Just goes to show that spam is often organised and it's usually worth looking deeper! And a reasonable question (though GTAIV requires re-opting-in on any wiki you've edited before, which is easier said than done when you don't read Urdu!). Different range to any of our friends above, but the range is blocked on enwiki (by me, for spam!). HJ Mitchell | Penny for your thoughts? 20:58, 23 March 2026 (UTC)
- Makes sense. If it helps, you can set the UI language in Special:GlobalPreferences to be English everywhere. lp0 on fire () 21:32, 23 March 2026 (UTC)
- Just goes to show that spam is often organised and it's usually worth looking deeper! And a reasonable question (though GTAIV requires re-opting-in on any wiki you've edited before, which is easier said than done when you don't read Urdu!). Different range to any of our friends above, but the range is blocked on enwiki (by me, for spam!). HJ Mitchell | Penny for your thoughts? 20:58, 23 March 2026 (UTC)
- That's genuinely a little surprising. Since CUs get GTAIV, could you find anything here on the same IP as the urwiki addition in January? lp0 on fire () 20:41, 23 March 2026 (UTC)
refspam bot
I would like to see a bot that would detect likely Reference spam, and generate a confidence score internally, the way User:Cluebot NG does. Ideally, at high levels of confidence, perhaps it could just revert as Cluebot does, but in any case, it ought to generate a project page with a table or log of rated edits so that humans could review the results, comment, perhaps define a confidence threshold for auto-reverts, and of course, provide data for refining and tuning the algorithm.
I seem to be spending more and more time analyzing and reverting WP:REFSPAM, and a lot of them are very obvious and really should not need human intervention. If someone is a new editor, adds substantially the same citation to multiple articles, with no added content (or brief, near-identical content), and has few or no edits outside one topic area (i.e., an WP:SPA), odds are very high they are a ref spammer. Mathglot (talk) 01:34, 24 February 2026 (UTC)
- @Mathglot: this is programmatically possible, a little difficult and time-consuming — but possible. I have a few questions, would it be okay if we continue the discussion on your or my talkpage? —usernamekiran (talk) 08:37, 23 March 2026 (UTC)
- usernamekiran, sure, let's move it, but let's find a more public Wikipedia page or WikiProject page where we have a prayer of attracting other interested comment. Perhaps Wikipedia talk:Spam (540 watchers, 56 pageviews/mo.) or Wikipedia talk:WikiProject Spam (1,218 / 1,926), or a subpage of one of them to centralize possibly extended commentary? My knowledge in this field is ancient now, but I wonder if we compiled a grab-bag of possible features (to add to the four I listed), generated a test set of a few hundred human-assessed spam evaluations, and threw a machine learning bot at it with the feature set, whether that might generate a usable model, at least as proof of concept. Likely with all the advances in AI, some of that can be streamlined, maybe even the assessments? That would be a win. Mathglot (talk) 09:19, 23 March 2026 (UTC)
- That's even better, I should have thought of that. I have a primary workflow in my mind, only for creating the report(s). In the early phase, the bot should rely on heuristics instead of machine learning. Once we create a good confidence scoring mechanism, we can move to next phase of reverting the edits. But during first phase, we will need inputs from other users on reports — to cross-verify the suspected spam links. In few hours, I will copy-paste this conversation, and detailed workflow on Wikipedia talk:WikiProject Spam, and notify few relevant venues of the discussion. Once we create a workflow/logic, I can start on concrete programming. During the discussion, I will create code for detecting the URLs being inserted, associating them with users/articles, and other basic necessary stuff. —usernamekiran (talk) 23:28, 23 March 2026 (UTC)
- usernamekiran, sure, let's move it, but let's find a more public Wikipedia page or WikiProject page where we have a prayer of attracting other interested comment. Perhaps Wikipedia talk:Spam (540 watchers, 56 pageviews/mo.) or Wikipedia talk:WikiProject Spam (1,218 / 1,926), or a subpage of one of them to centralize possibly extended commentary? My knowledge in this field is ancient now, but I wonder if we compiled a grab-bag of possible features (to add to the four I listed), generated a test set of a few hundred human-assessed spam evaluations, and threw a machine learning bot at it with the feature set, whether that might generate a usable model, at least as proof of concept. Likely with all the advances in AI, some of that can be streamlined, maybe even the assessments? That would be a win. Mathglot (talk) 09:19, 23 March 2026 (UTC)
- post move comment:
- I think we should start with a narrow first phase that only collects/processes data, and produces reports for human review. No automatic reverts would be attempted at this stage.The first version would focus only on detecting references that are added in edits and recording structured information about them. Technically, the bot would examine revision diffs (not full page text) to identify URLs that were newly inserted. For each detected URL it would record the editor, the article, the timestamp, and the domain of the link.From this dataset the bot can generate periodic reports showing patterns such as a user repeatedly adding the same domain across multiple articles.I am thinking about publishing the report to bot's userspace/subpage where editors could review cases and optionally mark them as “spam” or “not spam”. Those human labels/cross-verification could later be used to refine the detection heuristics.After the reporting system is working reliably, and a useful dataset has been gathered, we could improve the heuristics and introduce a better confidence score.Automated edits would only be considered after thorough testing and community review.In short, Phase 1 would simply detect reference insertions and produce review reports. Phase 2 could potentially introduce automated reverts. Once a rudimentary code-base/skeleton is created, I will post it to github. Any technical, and non technical suggestion/help would be appreciated a lot. After a little discussion here, I will also post a notification of this discussion at WP:VPT. courtesy ping to Mathglot —usernamekiran (talk) 02:28, 24 March 2026 (UTC)
- I created a base code for the bot successfully. It is not advanced yet, but it works. To improve it, I need more input data. For now, I need 1) a few accounts (no matter how old/stale/recent) that added same or different spam URLs to multiple articles. 2) any article where same spam URL was added/reverted in short period - like 7-8 hours (or maybe more) either by same or different accounts. From the top of this page I got Almas-7866, and Aryansrv. I will go through page history of this page for more. But if you come across anything, kindly post it on User:KiranBOT/REFSPAM/verified spam. —usernamekiran (talk) 08:13, 24 March 2026 (UTC)
- Have you thought about the possible overlap with COIBot? lp0 on fire () 09:49, 24 March 2026 (UTC)
- @Beetstra and Lp0 on fire: dang! That's the bot I was thinking of! I couldn't recall the name, I went through top 100 bots (I knew it was in top 100), saw COIBot but thought it was some other bot, later I thought I had made up the bot with the link task. COIBot already has great code, and access to other data. There's no point in creating another bot with same goal. It's like reinventing the wheel. If necessary, new functions can be added to COIBot. —usernamekiran (talk) 13:17, 24 March 2026 (UTC)
- I do think there's a good idea here. Personally, I would love something that flags all external URLs added by newish (say, no-EC) accounts that aren't to sites already linked (say) 100 times, so a new editor adding links to the New York Times won't be flagged but an account that used the suggested links feature a few dozen times and then added a spam link would be flagged. There would be false positives, but the rate should be manageable and we would hopefully catch spammers before they manage to spam the same link 60 times. HJ Mitchell | Penny for your thoughts? 16:19, 24 March 2026 (UTC)
- @Beetstra and Lp0 on fire: dang! That's the bot I was thinking of! I couldn't recall the name, I went through top 100 bots (I knew it was in top 100), saw COIBot but thought it was some other bot, later I thought I had made up the bot with the link task. COIBot already has great code, and access to other data. There's no point in creating another bot with same goal. It's like reinventing the wheel. If necessary, new functions can be added to COIBot. —usernamekiran (talk) 13:17, 24 March 2026 (UTC)
- Have you thought about the possible overlap with COIBot? lp0 on fire () 09:49, 24 March 2026 (UTC)