This page is used for discussions of the operations, technical issues, and policies of Wikimedia Commons. Recent sections with no replies for 7 days and sections tagged with {{Section resolved|1=--~~~~}} may be archived; for old discussions, see the archives; the latest archive is Commons:Village pump/Archive/2024/09.
Please note:
If you want to ask why unfree/non-commercial material is not allowed at Wikimedia Commons or if you want to suggest that allowing it would be a good thing, please do not comment here. It is probably pointless. One of Wikimedia Commons’ core principles is: "Only free content is allowed." This is a basic rule of the place, as inherent as the NPOV requirement on all Wikipedias.
Any answers you receive here are not legal advice and the responder cannot be held liable for them. If you have legal questions, we can try to help but our answers cannot replace those of a qualified professional (i.e. a lawyer).
Your question will be answered here; please check back regularly. Please do not leave your email address or other contact information, as this page is widely visible across the internet and you are liable to receive spam.
Purposes which do not meet the scope of this page:
SpBotarchives all sections tagged with {{Section resolved|1=~~~~}} after 1 day and sections whose most recent comment is older than 7 days.
September 11
Proposal: AI generated images must be clear they're AI in the file name
Now these are being used for a good purpose (reporting on the misuse of AI at en:Wikipedia:Signpost) so please don't just nominate them for deletion, but look at these filenames:
Now I've moved them to File:AI genetated image of... - but that's just actively setting people up to use semi-believable illustrations that have no scientific accuracy, and then making it relatively hard to catch what happened.
Should we be somewhat stricter about filenames for AI? There's cases where I think it matters less than these, but the capacity to mislead is high. Adam Cuerden (talk) 10:01, 11 September 2024 (UTC)[reply]
File names for AI generated images not indicating that's what they are is definitely an issue. There's no reason there shouldn't be some indication in the file name that an image is AI generated. I think it would be in alignment with the changes to guideline on how to name files that was passed recently to. Regardless, file names should be as descriptive as possible and I can't see why that information shouldn't be included in the file name. --Adamant1 (talk) 11:02, 11 September 2024 (UTC)[reply]
An alternative would be to add a tag to the media without requiring files to be moved or named with that in the title from the start. Just like any NSFW image has an indicator for that on sites like reddit. It would be shown on all files in Category:AI-generated images either at the end of the file-title or e.g. within a corner of the thumbnail. I think adding that automatically would be better. However, when uploading the file using the Upload Wizard and checking made with AI one could also automatically append (AI-generated) or (made using AI) to the file-title. Prototyperspective (talk) 11:29, 11 September 2024 (UTC)[reply]
The template only shows on the file page. And even there it doesn't look very different from other common license templates which people only interested in the content usually probably don't look at either and many files like the linked examples don't even have these templates. Prototyperspective (talk) 12:47, 11 September 2024 (UTC)[reply]
Support The file name and caption are the first things you see when you select an image that's used on Wikipedia for better viewing, right before you click through to the file's own page. For most people it'll probably be the only information they'll see. This information is absolutely important enough that it should be mentioned in the file name. ReneeWrites (talk) 13:03, 11 September 2024 (UTC)[reply]
Support per ReneeWrites. I agree that having AI generated marked in the file name will give Wikipedia users much more transparency on the provenance of files. William Graham (talk) 15:36, 11 September 2024 (UTC)[reply]
Oppose. Nice idea, but I prefer templates that can be translated and add properties. Latin letters in a filename are not a good clue in other scripts. We may have endless rename requests. File naming hacks are also not systematic; we do not routinely encode other properties in filenames. Glrx (talk) 15:47, 11 September 2024 (UTC)[reply]
At the same time, an actively misleading filename is a problem. They are not leucocytes (for example. They don't even look much like cells. AI is very good at creating images that look like they're plausible depictions but really aren't, they just ape the - for lack of a better word - art style of real scientific illustrations, coloured electron microscope depictions, and so on. Adam Cuerden (talk) 16:05, 11 September 2024 (UTC)[reply]
we do not routinely encode other properties in filenames - We routinely use naming systems like "Flag of [Country]" for other types of files. Using filenames to make important disclosures about the origin of files isn't a huge leap. Omphalographer (talk) 19:11, 11 September 2024 (UTC)[reply]
Support Update with caveat: my support is for this idea in principle, with the understanding that we would need an additional discussion about implementation to cover things like wording. — Rhododendritestalk | 12:32, 12 September 2024 (UTC) This is a good idea, and in line with the spirit of many off-wiki policies proposed for AI content. It also doesn't preclude a template. The question, though, is what label/language should be used. It would need to be something someone wouldn't choose accidentally for a non-AI image. Also, documentation for this rule would need to be clear that we're talking about media that is produced through generative AI models (as opposed to, say, a scientific visualization in which machine learning was used somewhere in the process). — Rhododendritestalk | 16:10, 11 September 2024 (UTC)[reply]
I wouldn't be too concerned about language necessarily. If the filename is not in a language people speak, they're much more likely to check the decription. We don't need a perfect solution, just an improvement. Adam Cuerden (talk) 16:14, 11 September 2024 (UTC)[reply]
Comment - This is a good idea, but it needs refinement. Besides Rhododendrites's caveat's above, I think it should only apply to images which depict something in a realistic manner. There's not much point requiring this for something like File:Portrait of a Unicorn.png. Otherwise, I would only support it as a recommendation, not a requirement. Nosferattus (talk) 18:21, 11 September 2024 (UTC)[reply]
We shouldn't be mass changing file titles without a compelling reason. It just generally causes chaos. I think files being factually misleading is a compelling reason, but otherwise, I'm not convinced. Nosferattus (talk) 16:57, 15 September 2024 (UTC)[reply]
I'd say there's a class of images where it matters less. But a human-made illustration probably has some effort to get key aspects, whatever those might be. AI just tries to get something that looks like other images with similar key words, and might miss out important bits that a human wouldn't. Honestly, as a general rule, the higher the likelihood it'd be used on Wikipedia, the more that's an issue. Adam Cuerden (talk) 19:34, 11 September 2024 (UTC)[reply]
Agreed on Nosferattus' conditional support: images that can be mistaken for something else, should be marked, and the filename is the most obvious place to do so. By the way, this also applies to photoshop fabrications of "real life flower elfs" etc. And from a filemover perspective: We are supposed to only rename files that are realistically going to be kept. Is there even a rationale to keep misleading non-scientific AI illustrations? I mean, beyond illustrating how you can't trust AI illustrations? --Enyavar (talk) 00:01, 13 September 2024 (UTC)[reply]
Support, as we should use any (and all) means to achieve maximum transparency for re-users about the non-authenticity of AI-generated images. --Túrelio (talk) 18:27, 11 September 2024 (UTC)[reply]
Support: I don't see any downside. One remark, though: like everything else on Commons, this should not be restricted to English, and I don't imagine I would recognize something if it were marked in Chinese as AI. How do we intend to deal with the multilingual aspect of this? - Jmabel ! talk19:32, 11 September 2024 (UTC)[reply]
As I said above, I think that perfection isn't needed. If it's labelled in Chinese, as long as the whole filename is in Chinese, Anglosphere people will presumably go to the description. They might not for one that has a plausible English filename. Adam Cuerden (talk) 19:37, 11 September 2024 (UTC)[reply]
Yep. In a Czech file name, the warning should be in Czech, and in a Japanese file name, in Japanese - tailored to the native languages these images are likely to get used for. And if I'm that determined to use a cool image with Tamil filename in the German WP, I the user must make sure to understand the filename and description. (GTranslate exists.) --Enyavar (talk) 00:10, 13 September 2024 (UTC)[reply]
Comment from a filemover. If you want to apply this requirement to files after upload, you should amend Commons:File renaming to make it clear that lacking a statement of AI-generation in the filename is good cause for renaming. Either by adding a new numbered criterion or by finding a way to shoehorn it into an existing one (2 or 3, I'd guess). --bjh21 (talk) 21:01, 11 September 2024 (UTC)[reply]
@Bjh21: You could argue, in clear cases like the ones I mentioned earlier, it's already covered by 2, since they aren't actually pictures of (say) leukocyctes, but I agree that adding an example would help. Adam Cuerden (talk) 20:40, 12 September 2024 (UTC)[reply]
It'd be nice to expand criterion 2 to allow adding information about the non-factual nature of an image in general (e.g. AI generated images, simulations, reenactments, historical reconstructions, artistic representations, etc). Omphalographer (talk) 22:40, 13 September 2024 (UTC)[reply]
I believe that ALL substantially manipulated or fabricated images, that might be misunderstood as "real", should be very clearly marked. It shouldn't depend only on whether someone decides or rules that the fabrication method was "AI". ITookSomePhotos (talk) 21:37, 11 September 2024 (UTC)[reply]
Support AI generated images must have "AI" in the file name as a principle, perhaps even better would be "AI generated", which is more clear. And always at least in Latin letters. Yes, the backlog might be a problem, but we can start now for new uploads. --JopkeB (talk) 05:12, 12 September 2024 (UTC)[reply]
What do people think about when uploading the file using the Upload Wizard and checking made with AI one could also automatically append (AI-generated) or (made using AI) to the file-title? (No replies on that above or on the idea of a tag displayed dynamically next to the file-title and in the thumbnail.) I think doing something automatically and in a standardized way would be better than just requiring this which many uploaders will not follow up on. Prototyperspective (talk) 10:49, 12 September 2024 (UTC)[reply]
Given that the Upload Wizard already asks about AI tools, I think it would be appropriate for it to ensure that uploads using them follow whatever policy arises from this discussion. --bjh21 (talk) 17:26, 12 September 2024 (UTC)[reply]
Support. on top of this, if we need to rename these files, i suggest requiring the new name to begin with " «AI generated» " or " ~AI generated ". this will make them appear behind all ascii letters when sorted alphabetically. RZuo (talk) 13:58, 12 September 2024 (UTC)[reply]
If you want to change where something sorts, I think it's better to do it using {{DEFAULTSORT}} rather than by requiring a particular pattern in the filename. --bjh21 (talk) 15:46, 12 September 2024 (UTC)[reply]
Please no. This makes file names unnecessarily difficult to type - most keyboards don't have «» keys, and ~ is difficult to find on many mobile devices. The goal is to label these files, not to make them difficult to use. Omphalographer (talk) 22:53, 12 September 2024 (UTC)[reply]
It really depends on the meaning of "Special", lest we ban, say, Korean file names, or accents. We have French filenames with French-style quotes in them, and we shouldn't change those. At the same time, we have default sort; let's not make it a policy to name AI images File:💩AI generated💩 Foobar.jpgAdam Cuerden (talk) 10:03, 13 September 2024 (UTC)[reply]
That's not really what I'm talking about. I don't think arbitrary putting brackets in file names is useful though. Maybe circle brackets, but «» or ~. If for no other reason then most keyboards don't have them to begin with. I'm also super annoyed by file names with emojis them though. They should 100% be banned. I'd be totally fine with requiring people put (AI generated image) at the end of a file name though. --Adamant1 (talk) 10:50, 13 September 2024 (UTC)[reply]
Aye, merely trying to avoid bad policy coming out of this. Should we append characters at the start of filenames to deprioritise them? No. That's a job for {{DEFAULTSORT}}. But «» are the standard quotes used in French, so we shouldn't ban their use, lest we require bad French. I'm a little bit of a stickler for trying to avoid policy for one situation that screws up other situations. Adam Cuerden (talk) 15:38, 14 September 2024 (UTC)[reply]
Keep it simple. Just put "AI" infront of the filename. Those who want to know more can check the summary / category of a file for details. Alexpl (talk) 11:06, 13 September 2024 (UTC)[reply]
Please rather put it at the end of the filename. Moreover, "AI" is ambiguous and also included in many other files, so again I'd suggest (AI-generated) or (made using AI) and this could be appended to the initial file-titles automatically in the Upload Wizard. Prototyperspective (talk) 11:24, 13 September 2024 (UTC)[reply]
the thing is, if you prepend filenames with A, it's counterproductive to your aim (discouraging use of ai files as illustrations) because then all the ai files will occupy the front rows in categories (unless you add defaultsort of a super "late" unicode to the ai template). RZuo (talk) 18:34, 16 September 2024 (UTC)[reply]
The examples people have brought up in support or as examples in this discussion include "AI generated", not just AI. Edit: Except for Alexpl's comment. ReneeWrites (talk) 14:22, 13 September 2024 (UTC)[reply]
Support When images are used internally editors know to check the filepage, when outside people reuse the images they are usually not aware of the info on the files page. We have had people reusing images without complying to the terms of the license by naming the creator. I would like to see (AI generated) the way we use "(cropped)" in file names. --RAN (talk) 18:06, 13 September 2024 (UTC)[reply]
Ooh, that's a good point - if a user saves an image from Commons (or another wiki) to their computer, for instance, there's a decent chance that its file name will remain intact as a warning. A template on the file's talk page certainly won't. Omphalographer (talk) 18:13, 13 September 2024 (UTC)[reply]
Strong oppose no no no, file names shouldn't be used for that. This opens up Pandora's box for controversial subjects. This should be in the file description, not in the name. Multichill (talk) 12:53, 15 September 2024 (UTC)[reply]
Comment Why do we even accept AI generated images to begin with? Most of them are misleading, useless for Wikipedia articles, fake-y look as standalone content, and can be barely trusted. AI images should only be limited to very specific scenarios, otherwise we end up with a bunch of superheros holding the Commons logo, which we can all agree is largely a set of very interesting trademark violations and not consistent with community practices. Moreover, there have been recently a number of court cases around copyright infringement for several of these AI companies, so I'm concerned that we can't distinguish the provenance from different models that may or may not be trained on infringing datasets. We have no idea how this is going to be regulated.
Comment I generally think templates and/or structured data are a better way to do this, and more multilingual-friendly. But I'm not opposed to also requiring it in file names if the community agrees on that (and I will anyway include in the filename of any AI files I upload myself). Noting that I have made a related proposal below: #Proposal: de-prioritise AI images in searchthe wub"?!"15:09, 19 September 2024 (UTC)[reply]
Support this would prevent problems with AI images being mistaken for real ones, and I really don’t see how this could potentially be harmful. Dronebogus (talk) 11:13, 24 September 2024 (UTC)[reply]
I'm sensing pretty widespread support, so let's plan out what would need changed:
Commons:File renaming: #2 gains "To identify AI generated works" with a possible more general version of "to point out major manipulations" (colourization, etc). This is explicitly allowed to be in any language.
Commons:AI-generated media notes that the AI-generation must be mentioned in the filename, ideally in the same language as the rest of the filename.
File upload wizard appends "AI generated" if the AI creation option is ticked, with the option to change this after, but with a note saying that identifying AI art in filenames is important. Alternatively, this can just be a soft prompt, that suggests a new filename, but doesn't require. (Similar to others where you can click "ignore and upload file anyway)
I think #4 should either not be implemented or be for files in Category:AI misgeneration. Images shouldn't be sorted by how they were produced but by by where the user is expecting to find them / looking for them or generally the relevance and quality of the image as it relates to the category concept, not the method/techniques used to produce it. You may have missed an addition to Commons:File naming. Prototyperspective (talk) 20:40, 14 September 2024 (UTC)[reply]
Added File naming, and you're probably right about #4. Wanted to pull all the suggestions made, but that may be too much (if nothing else, AI image categories wouldn't do the headers for first letter of filename). Adam Cuerden (talk) 11:44, 15 September 2024 (UTC)[reply]
#3 and #4 are terrible ideas. #3 will cause uploader confusion, filename conflicts, language issues, etc. This needs to be done by humans, not machines. #4 will also be confusing as no one will expect {{PD-algorithm}} to mess with the sorting. Plus it's just unneeded and potentially unhelpful, as there may be other reasons an AI-generated file needs to be sorted in a particular way. Nosferattus (talk) 17:05, 15 September 2024 (UTC)[reply]
I don't think they're terrible ideas, but I don't think these two are needed. We're not inundated with such a flood of AI-generated images being uploaded to Commons that these couldn't be done by hand, and a lot more people have filemover rights than admin rights, so this wouldn't add to the backlog of issues needing admin attention. ReneeWrites (talk) 09:46, 16 September 2024 (UTC)[reply]
Okay. Then let's focus on points 1, 1b, and 2. For 1b, I'm thinking (under "Clear")
"Where an image, either through method of creation or modifications, might mislead, this should be noted in the filename. This includes AI generation, colourization of a photograph, turning a sepia image black and white, upscaling an image, and other things that might not be immediately obvious. Simple, minor fixes do not need to be noted."
This discussion is about AI, and I think we should stick with that. I've seen way too many discussions get killed the moment they gain any traction because people keep attaching stuff to it that is tangentially related that no consensus was reached on. ReneeWrites (talk) 15:22, 16 September 2024 (UTC)[reply]
1a: Support "To identify AI generated works" sounds good to me.
1b: Comment I think it would be more at home under "Descriptive", specifically the subheader "Correct". There's nothing particularly unclear about the filename "Cancer cell.jpg", but it leaves out a lot of pretty crucial context that makes it pretty misleading.
I'd like to propose this change: Correct – The name should describe the file's content and convey what the subject is actually called. Inaccurate names for the file subject, although they may be common, should be avoided. The title given to a work of art by the artist that created it is considered appropriate, even if the name has nothing to do with what is depicted (for example, many works of Dadaism). The name should also be free of obvious errors, such as misspelled proper nouns, incorrect dates, and misidentified objects or organisms. Users are allowed to upload "unidentified" or "unknown" organisms but such files may be renamed upon identification. AI-generated images must disclose this fact in the file name.
It's tempting to include a bit on the rationale as to why, but none of the other examples have that either, they simply state what is policy. So I think addressing this with just one line that's clear and unambiguous is both pragmatic and in line with how the rest is written. ReneeWrites (talk) 15:50, 16 September 2024 (UTC)[reply]
Made a slight adjustment to the wording. A lot of the file names on Commons made with Dall-E or Midjourney have that in their file name, which should also cover this base. ReneeWrites (talk) 08:42, 17 September 2024 (UTC)[reply]
Okay. If no-one else has suggestions after a couple days, let's bring 1, 1b (with your text), and 2 together, ping everyone involved in the original discussion, and implement. Secondary ideas can be considered after that. Adam Cuerden (talk) 17:21, 17 September 2024 (UTC)[reply]
May be worth a clear call on whether AI upscaled photos should fall under "AI generated" for all this, given their similar potential for being misleading when the viewer doesn't realise that an AI was involved (eg. File:2Pac Passport (cropped).jpg, where one Wikipedia editor was pleased to find what they described as a "free-use authentic high quality photograph" of the subject on Commons, but no, it's just an upscale of an old and extremely low quality passport photo). Belbury (talk) 12:58, 25 September 2024 (UTC)[reply]
It's not AI-generated, it's AI-upscaled which is very different. It needs separate templates and categories which also warn the user about issues like potential inaccuracies. Prototyperspective (talk) 13:35, 25 September 2024 (UTC)[reply]
It's different, but if there's going to be a policy change on naming and negative-boosting AI content, we should be clear whether that also applies to AI upscaling or whether it doesn't apply to it at all. Some users already (very understandably) tick "I generated this work using an artificial intelligence tool" when uploading an AI-upscaled image, causing it to be incorrectly filed as {{PD-algorithm}} with no human authorship. Belbury (talk) 14:17, 25 September 2024 (UTC)[reply]
I think when this is checked the Upload Wizard should show another checkbox about whether img2img (an input image) was used or whether upscaling was used. If the former is checked, the user should enter some url to the input image(s). If the latter, it would add the template for AI upscaled image. Prototyperspective (talk) 14:21, 25 September 2024 (UTC)[reply]
Upload wizards capacity
I note in this AI discussion that the upload wizard asks the question of is it ai generated, so its possible that appending to file names or adding a template to identify AI generated media could be relatively easy to do automatically at upload. with a high degreee of consistance Gnangarra01:06, 17 September 2024 (UTC)[reply]
{{PD-algorithm}} is already added to any upload where this box is ticked, in addition to the licence template specified by the uploader.
It's worth remembering that some users tick this box in error, fairly regularly. Any additional effects of ticking it will require additional steps of cleanup in that minority of cases. Belbury (talk) 13:46, 19 September 2024 (UTC)[reply]
@DarwIn I already relayed the idea of adding automatically a template, if I remember correctly something is already added. Anyway, if there is community consensus to add a(nother) specific template, I can relay this too and discuss it with the team. It's going to take some time anyway. Sannita (WMF) (talk) 13:47, 19 September 2024 (UTC)[reply]
However it should be possible to de-prioritise such images by adding Template:PD-algorithm (which all AI generated images are supposed to be tagged with) to MediaWiki:Cirrussearch-boost-templates. I would suggest a ranking of 50%, which would rank AI files below most other files, but above those nominated for deletion or otherwise tagged as low quality/superseded.
Example of a search for "animal" (cluttered with many old book scans crowding out already badly prioritized images)
Strong oppose Images made using some AI tools should either not be deboosted or only files in Category:AI misgeneration should be sorted to be further down in general: images shouldn't be sorted by how or by which techniques and tools they were produced but by where the user is expecting to find them / looking for them or generally the relevance and quality of the image as it relates to the category concept, not the method/techniques used to produce it. Additionally, this functionality only incentivizes people to hide the fact they used AI to make an image which we should encourage people to specify. There are many cases where AI images are some of the best for the concept searched for and they should be shown fairly high up in such cases, especially when they're in use. The example search results look like that because nearly all images for a useful Wikibook used rhino images. They indeed should not show high up there (for this search term) but that doesn't mean all AI images should be downranked in general. Some other improvements to the search results should be made that are not specific to AI...for example only showing used images high up if they are used on Wimedia items (like Wikipedia articles) that are actually related to the search term. I will attach a screenshot that shows how bad the search results often are, this is a general issue and lots of other media crowd out useful media. Some other issues include that it often shows very outdated charts at the top when there are more up-to-date charts. This is just some discrimination of the use of a novel tool to produce often high-quality useful images but not a reasonable effective measure to actually improve the search results. Prototyperspective (talk) 15:06, 19 September 2024 (UTC) Added screenshot --Prototyperspective (talk) 15:27, 19 September 2024 (UTC)[reply]
" or only files in Category:AI misgeneration should be sorted to be further down in general" I added that a whole month ago already Trade (talk) 18:25, 20 September 2024 (UTC)[reply]
Instead we could have a filter that excludes AI images. Moreover, these aren't a problem here the example given is one of very very few cases where the results are indeed cluttered by AI images. Third, again the tools and techniques used for producing an image is not the important part...rather whether it's high-quality, useful and relevant to the search term. I don't see this problem on Image search engines but I also don't know what (and how) you searched for. What's next – downranking images made or edited with Photoshop or Krita?Prototyperspective (talk) 11:00, 20 September 2024 (UTC)[reply]
Neutral I'm not sure this is the answer to dealing with the problems caused by AI images. I think another proposal for putting "AI generated" in file names of AI generated artwork would be a better solution. Since it's not so much where the images show up in search results, but that people often don't know the images are AI generated to begin with. People can still accidently get an AI generated image if they are pushed down in the search results. What we need is better ways to clearly identify and deal with AI generated artwork to begin with. I feel like it's mostly a curation issue at this point though. --Adamant1 (talk) 00:32, 20 September 2024 (UTC)[reply]
Why discriminate against the use of one particular tool? I don't see any rationale there only that you apparently want to continue to subjectively semi-censor images made with a tool you apparently don't like. Prototyperspective (talk) 11:02, 20 September 2024 (UTC)[reply]
AI generated artwork isn't a tool, its a specific type of art. Stop trying to act like we're just discussing Microsoft Paint every time this comes up. Its getting rather tendentious. --Adamant1 (talk) 11:08, 20 September 2024 (UTC)[reply]
It's the AI tool. One can use the result and edit it in Krita/Photoshop or create a draft and use an AI tool to improve it via img2img. AI-generated art is a type of art that makes use of these tools, similar to photography by definition makes use of cameras or paintings characteristically make use of paintbrushes. The level of involvement of the human is often far lower than in manual paintings. I doubt people voting here have much experience with prompting around for a few hours to get an image to look nearly exactly like you want it to or seen cases where AI images are the most illustrative. Stop trying to act like it's not discrimination against one particular novel tool (or art-technique/artform). Prototyperspective (talk) 11:18, 20 September 2024 (UTC)[reply]
Don't get me wrong, I'm not acting like it isn't singling out a specific type of art. It has to be due to the nature of the thing. "Artwork" isn't the problem here. AI generated artwork is. You know as well as do that no one is uploading personal drawings of cells made in Photoshop and trying to pass them off as actual images. Let alone that anyone will be fooled by said images like they clearly are for AI generated artwork. Your just being disingenuous and choosing to ignore the issue instead of admiting that AI artwork has its own problems that inherently don't exist with other artforms. --Adamant1 (talk) 11:31, 20 September 2024 (UTC)[reply]
drawings of cells made in Photoshop and trying to pass them off as actual images Delete these. There is an ongoing proposal about requiring their file titles to specify they are AI-generated and that seems like not a bad idea. There also are templates and categories by which people can see that the file is AI made even if that's not in the file description where this info should be and usually is. You didn't address the points made but if that helps: yes AI artwork have their own problems and I never said otherwise. One could even think about downranking AI images for search terms where the user likely looks for actual depictions and not artwork. For example, when searching for "animals" all art media and AI-generated media could be downranked and images featured in WP articles broadly about animals like such as the collage images upranked. Prototyperspective (talk) 11:53, 21 September 2024 (UTC)[reply]
Comment I'm not sure if nobody else has noticed this or people have noticed but simply haven't commented, but the vast majority of the Javan rhinoceroses generated by the AI prompts aren't Javan rhinoceroses, but black or white rhinoceroses. ReneeWrites (talk) 08:38, 21 September 2024 (UTC)[reply]
The search results for that are also littered with AI generated images (Rhinoceros sondaicus). And search takes multiple things into account (descriptions, file names, metadata) so searching for both should yield similar results. One is not more correct than the other. ReneeWrites (talk) 10:50, 21 September 2024 (UTC)[reply]
Doesn't that just illustrate a quirk of MediaSearch that displays images not categorized as "Rhinoceros sondaicus" also ? Compare with Special:Search linked above.
In general, we might be blaming "AI" for defects of our search engine due to our naming conventions for categories and files about "Javan rhinoceroses". The correct approach would be to fix the search engine for the general issue. ∞∞ Enhancing999 (talk) 11:06, 21 September 2024 (UTC)[reply]
They are categorized as rhinoceros sondaicus, specifically Category:Rhinoceros sondaicus in art. Thinking on this a bit more, I don't think de-prioritizing search results would fix the issue at hand. These are simply not images of Javan rhinoceroses, even if the prompt called for those to be generated. ReneeWrites (talk) 16:38, 21 September 2024 (UTC)[reply]
Comment Also please consider all the unintended consequences like if a logo of some project or tool was made using AI (as is the case for several Wikimedia projects) it wouldn't be shown at the top when searching for the project name even if it's the actual logo (and this was just one example). Also before considering this please consider that artistic and illustration workflows are involving to substantially incorporate AI-generation and we'd be indiscriminately semi-censoring lots of media (that at times is quite high-quality or relevant to whatever the user searched for) – so please see this example workflow video of an AI tool integrated into Photoshop. I think it should probably be required watching especially for people who don't have a lot of experience with using these tools and and in their personal experience didn't come across many media created using them that could be useful. Prototyperspective (talk) 12:12, 21 September 2024 (UTC)[reply]
Why don't you and Jeff include any explanation and don't address any of the points raised? What about the issue of it hiding the logos of projects etc when searching for these? Why are you suddenly in favor of indiscriminate (semi) censorship – I'd like to understand and think decisions should be based on sound well-thought through rationales. Prototyperspective (talk) 22:09, 21 September 2024 (UTC)[reply]
If some illustration uses some AI images because there are no good free ones it would get censored. If people make media in relation to politics we'd be be doing political (semi) censorship like China seems to do it – it's a big social problem if a large proportion of diverse media on all sorts of things is suddenly indiscriminately semicensored here or anywhere else. I doubt you have seen the video I linked and neither replying nor providing any rationale or addressing any points shows how this is about ignorance, not anything thought through and possibly some emotional knee-jerk reaction because voters simply don't like the use of AI tools for whatever personal reasons. I think Wikimedia projects are generally oriented towards rationality and reason, not subjective opinion vote counts. The AI images people have uploaded here so far are below average but that may change and there's many cases where these are very relevant for various search terms.
@Borvan53: you used an AI tool to create what seems to be the first and high-quality image of a palantir which is a trope of fantasy fiction and used in several Wikipedia articles – what do you think about this proposed policy that would hide your image from the search results even when searching for "palantir"? @TatjanaClimate and S. Perquin: @Milena Milenkovic (VMRS), Fuzheado, and JPxG: you created some logos using AI tools, what do you think of this proposal that would hide in the search results such when searching for the name of the project/… for which the logo is used or related search terms? @Raresvent: An AI image of your prompting is a quite good illustration of post-apocalyptic art, what do you think of moving it far down in the search results for that search term? Prototyperspective (talk) 23:33, 22 September 2024 (UTC)[reply]
Oppose There is already the template {{PD-algorithm}}. It is mandatory. AI images will be more and more common, and our thinking must be focused about their quality. Should I mention that I worked a couple of hour to generate the image of the Palantir ? First trials gave everything but a Palantir. Borvan53 (talk) 06:51, 23 September 2024 (UTC)[reply]
I was not saying you need one, just asking for one. Moreover, the linked policy suggests when discussing proposals people if possible shouldn't just leave a vote to be counted but also e.g. address points raised.
@Adamant1: , I am allowed to ask relevant people to participate in relevant discussions and lots of people did this elsewhere on WMC. I think comments that do not address any critical points & questions and have no rationale should be ignored, you think people affected by this proposal should be ignored so we disagree on that. Most contributors do not check VP here, which is frequented by a very small fraction of users, but also most users probably don't want AI images to be semi-censored or would like to have policy-discussions to be based on sound reasoning. These users may provide some further insights and I made transparent how they are related to it. If that wasn't okay then where is a policy that prohibits doing so and why can other users do this over and over in other discussions without any issues? Prototyperspective (talk) 08:53, 23 September 2024 (UTC)[reply]
Sure, Perspective. There's a difference between that and pinging specific people who you know will take your position because they have uploaded AI generated artwork before though. Otherwise I could give a crap, but if your ping people who are clearly going to have a particular position about it then 100% your just canvasing at that point. Especially since you did it half way through the discussion and when there was more support for it then not. So it's pretty obvious that's what you were doing. --Adamant1 (talk) 08:59, 23 September 2024 (UTC)[reply]
Again, I never seen any policy that prohibits that which may be because Wikimedia was originally meant to have discussions based on reasons rather than votecounts but if one such exists please link it and if not you could propose one. Secondly, again many other users including admins did the same in other discussions such as DRs and they didn't transparently show how these users relate to the subject which is my third point: I made this transparent so everyone can contextualize their comments, explanations, and votes. Affected good-faith long-standing users are I think certainly users that should be involved in proposal discussions and we may disagree on that. I also think in politics that people who are affected by policy decisions should have a say and we may disagree on that as well. Fifth, you and others could also invite users but I think the small number of users who frequent this VP board have much of an overlap with the strongest opponents of AI tools in media production. It may indeed be a problem if I did that early in the discussion rather later one or without making their relation to the subject transparent. Prototyperspective (talk) 09:36, 23 September 2024 (UTC)[reply]
Comment Take the example of the project of biological expertise/palaeontology that takes bones of early humans or dinosaurs and tries to add muscles, tendons and hair to figure out what the actual human/animal looked like. Are the results in the category of AI generated images? - R. J. Mathar (talk) 19:05, 22 September 2024 (UTC)[reply]
I don't know if deprioritizing them across all searches the best way to do this, but I think there absolutely has to be some way to choose to exclude them from a search. JPxG (talk) 23:47, 22 September 2024 (UTC)[reply]
A filter for AI images is a separate proposalA filter the user can enable to exclude AI media is a separate idea and was proposed by the same user elsewhere. I think I would support that as well and it's one reason why I put so much effort into categorizing AI images into the AI-generated media subcats which should now contain virtually all of these. Prototyperspective (talk) 23:52, 22 September 2024 (UTC)[reply]
Oppose I don’t particularly care for AI art; but AI art is not overwhelming Commons by a long shot. Often times AI is used for illustrating fanciful concepts, where it is not inherently better or worse than human art. And it’s highly unlikely to get ranked highly among depictions of mundane subjects that likely have featured, valued, or good images. So this, to me, reads like “AI is bad but we can’t ban it (at least not yet) so let’s quietly suppress it”. It feels kind of sneaky and manipulative, like Generic Big Tech Inc. manipulating what users see for generic sinister purposes. Dronebogus (talk) 12:16, 24 September 2024 (UTC)[reply]
I noticed recently that there's a special icon that can be added to valued images when their being used in galleries. It wouldn't help with this particular issues per se, but maybe having a little icon in the corner of every image that is AI generated so people will at least know that's what their looking at would help. It should be that hard to implement either. Although I suspect people could and would make the same arguments against something like that they normally do. But whatever. There should be something to indicate an image was generated by AI. The trick is to make it as inconspicuous as possible while still making it clear enough that people will notice it. --Adamant1 (talk) 12:23, 24 September 2024 (UTC)[reply]
advocate for nudism - not true: It is a protest in support of human rights as expressed in the UN declaration of human rights, the Istanbul convention, German constitution and German AGG law. And there is no nudity.
disturbing or offending - yes, there are people offended by the thought of equal rights for man and women
not educational - what? of course it is educational
So: when did Commons change it rules to allow only public domain content on the main page. Why are depictions of protests in support of human rights not allowed? and in what way is such a depiction not educational?
you could at least film with a tripod or a gimbal, and cut out the irrelevant parts like the first 1 min of this video, then it's something more worthy. RZuo (talk) 18:02, 21 September 2024 (UTC)[reply]
Seconding this. While MOTD, unlike POTD, doesn't require that content be featured (in large part because Commons:Featured media is not a particularly active process!), it should go without saying that files selected for MOTD should be of exceptional quality and should have self-evident educational value. This file isn't up to that standard. Omphalographer (talk) 23:41, 21 September 2024 (UTC)[reply]
I meant that apparently it's also not CCBY (public domain in the broader sense)
It does advocate for nudism, in this case for women to show their bare breast. I did not say there is nudity.
disturbing or offending was not the reason in itself. For very controversial files things should be in a neutral and reasonable manner when put onto the Main page. This is not a place to advocate your 'free the tits' opinions or however one calls it and this applies to other subjects but especially when things are quite controversial which results in people being repelled from WMC. Moreover, it is not about equal rights – seeing bare breast has generally different effects depending on sex, that is a neurological fact and of course varies with sexuality and there probably are also women turned on a lot by bare men's breast etc.
This is not a protest about in support of human rights even if you think or claim so. Nothing about is explanatory or truly educational. It does document an event and is slightly educational but it does not belong on the Main page. It's also not featured media. In which way is that educational? Compare this to videos explaining notable problems such as challenges facing the world's oceans or how something works or high-quality video of ecosystems/nature etc.
It's licensed CC-BY-SA-4.0, which is an acceptable license for both Commons generally and the front page specifically. It's not in the public domain, but files don't have to be in the public domain for them to be eligible for MOTD. It just so happens that a lot of media files that end up front-paged are PD. Everything else is a matter of opinion. The Free the nipple campaign is a legitimate campaign within the broader modern feminist movement, regardless of how you feel about the points it tries to get across or the methods they employ in doing so.
That being said, I do agree with RZuo's arguments that the media file itself is not of very high quality (footage is not stabilized, and the video was not edited to cut out the footage where nothing is really happening). ReneeWrites (talk) 18:16, 21 September 2024 (UTC)[reply]
I strongly disagreee with "footage where nothing is really happening". Of course there is. And no it does not "advocate for nudism". About "frequent obstrution of cars": It would falsify the situation to remove the cars. The file has a problem: audio and video are out of sync, but I am fixing that, i will upload a fixed revision by 18th. C.Suthorn (@Life_is@no-pony.farm - p7.ee/p) (talk) 19:15, 21 September 2024 (UTC)[reply]
I was referring to the template that says This image is not public domain. Please respect the copyright protection. It may only be used according to the rules mentioned here (or with an individual permission by the creator). This specifically excludes use in social media, if applicable terms of the licenses listed here not appropriate. so it seems like there are some additional restrictions, this may be wrong so maybe strike point 1 of 4.
The Free the nipple campaign is a legitimate campaign within the broader modern feminist movement, regardless of how you feel about the points it tries to get across or the methods they employ in doing so And? Are we now including activist videos of any type on the Main page so editors can advocate any bizarre fringe ideas they have on the very frontpage with little to no educational value etc? It doesn't belong there for many reasons and I elaborated on why while trying to keep it short. Prototyperspective (talk) 19:21, 21 September 2024 (UTC)[reply]
@Jeff G.: Not to speak for Prototyperspective but by fringe I assume they meant that the whole "free the nipple" thing is on the border or outer edges of the feminist movement. Which for all intents and purposes is the case. "Freeing the nipple" as it were isn't a part of regular, every day feminist discourse. Breast feeding in public maybe, but then it's not really inherently about "freeing the nipple" anyway. --Adamant1 (talk) 10:50, 22 September 2024 (UTC)[reply]
I suspect there's a misunderstanding around "This specifically excludes use in social media". Rather than adding additional restrictions for social media, I believe they're providing less restrictive attribution requirements for social media; on the linked page it says "At Social Media Sites only it is sufficient to use only the QR code inserted in the media file or near the associated media file". @C.Suthorn: can you confirm? Consigned (talk) 10:55, 22 September 2024 (UTC)[reply]
Indeed, on Social Media I offer this option for an easier attribution, that I am OK with. The template on the page on the other hand is only a clarification, because there are people who think, that everything on the internet is in the public domain. The same text (template) is also used by other users. C.Suthorn (@Life_is@no-pony.farm - p7.ee/p) (talk) 08:12, 24 September 2024 (UTC)[reply]
I'm not going to weigh into the whole thing about if a nipple campaign is a fringe idea or not since I think it's just bait for concern trolling on C.Suthorn's end and doesn't really matter in the grand scheme of things anyway, but I totally agree with RZuo and ReneeWrites about the low quality of the video. At the end of the day the frontpage should only have media that fits a certain standard and the video clearly doesn't meet it. I suggest this just be left there instead of needlessly turning it into an off-topic debate about what makes something "fringe" or not. --Adamant1 (talk) 05:18, 22 September 2024 (UTC)[reply]
Low-quality shaky cam video like this would only qualify for MOTD if it showed something momentous. This doesn't. I'm pretty comfortable in saying that at least 90% of what I've uploaded at Category:Videos by Joe Mabel is higher quality than that, maybe 100%, and I'm honestly not sure whether any of what I've uploaded would be good enough for me to advocated it for MOTD. - Jmabel ! talk21:40, 22 September 2024 (UTC)[reply]
@Prototyperspective: no, free licensed content is not "public domain in the broader sense". It is precisely "free-licensed". In particular, in countries where there is no requirement to attribute public-domain content, you can still be sued for using a CC-BY file without attribution. - Jmabel ! talk21:40, 22 September 2024 (UTC)[reply]
1. I don't think men with gynecomastia have the same or similar psychological-neurological effects on men as women's bare breasts on men. 2. I think they already do but this is somewhat offtopic: 3. This is not about whether the file is on WMC or whether that activism and its methodology make sense but whether such a file belongs on the Main page. 4. Jmabel, you're right...public domain is not the right term for what I meant with "public domain" there even when broadly speaking. Prototyperspective (talk) 12:13, 23 September 2024 (UTC)[reply]
Having a CC-BY-SA license is not a justifiable reason for removing content on the Main Page. It's a free license. Just wanted to address the original statement so that people who read this will see an answer without the hullaballoo about of the content of the video. Bastique☎ appelez-moi!21:37, 23 September 2024 (UTC)[reply]
(random headline)
Demonstrations and protests often take place on a regular basis, and this probably also applies to the Free The Nipple demonstration in Berlin at the Brandenburg Gate. So it is only a matter of time before there is a video that meets the technical requirements expressed here. I consider the issue to be an important one (I have already pointed out that the UN, EU and Germany are clearly in favour of equal rights) and reject the arguments ofdisturbing, offensive, nudism and fringe. So if there is a video with the desired technical requirements, I will post it on MOTD. If there is a formalised procedure at the time that applies to all MOTD (for example, a vote in which there are more approvals than rejections), I will of course adhere to it (and I am already curious whether there will still be audio files in OGG format with national anthems of micronations played listlessly by a US military band on the main page, where the lyrics are not sung, nor included as subtitles, on the file description page, or in the article about the national anthem, and the article also does not explain the history and meaning of the anthem (I.E. there are no double standards applied.)). --C.Suthorn (@Life_is@no-pony.farm - p7.ee/p) (talk) 08:39, 24 September 2024 (UTC)[reply]
These audio files were added by a user who is not blocked. They could be removed whenever they are replaced by something else. Yann (talk) 09:15, 24 September 2024 (UTC)[reply]
i dont think these low quality videos of unimportant events are comparable to music played by a professional military band. RZuo (talk) 14:17, 24 September 2024 (UTC)[reply]
I don't think recordings of scripted material that are inherently repeatable and uploaded from elsewhere are nearly as important as user-made videos of modern events.--Prosfilaes (talk) 17:50, 24 September 2024 (UTC)[reply]
Do we accept category pages which end with text like "Should you choose to perform actions before asking questions, your chances of getting detailed and polite answers will not be high."?
Regarding the 1st question: IMO they don't. The page should be modified or rewritten w/o personal attacks like "vandals", or be deleted (surely they can then recreate it on Russian WV, where there are no guidelines). --A.Savin21:49, 21 September 2024 (UTC)[reply]
I'd just delete both. About the page, the whole idea that someone should notify them before nominating an image of a monument for deletion is just laughable on it's face. The same goes for most of the other content on there. With the template, it seems to add a warning to everything it's added to saying "this template and pages using it are maintained by the Russian WLM team. Please read the guidelines before making any changes that can affect the monuments database!" Which is at least overly confrontational, if not totally pointless. It appears that their whole numbering system for "cultural heritage monuments" is a personal thing created by the Russian WLM team to. So I don't think it should be associated with the categories, images, or anything else on Commons. Especially if it's just going to be used as a way for Russian users to control things related to Russia on here. --Adamant1 (talk) 05:38, 22 September 2024 (UTC)[reply]
Well, this is the only database for Russian monuments. Nothing else exists at this level. Deleting the template would mean either (i) a lot of uploads of files without categories which nobody would notice; or (ii) we just stop WLM, which is the only source of files for Commons for most of these monuments. In the ideal world, the users would categorize the files properly, or it would be freedom of panorama for monuments in Russia, or Russia would become a civilized country and remove all these monuments to Lenin etc from the protection lists, but none of this is likely to happen in our lifetime. Ymblanter (talk) 06:27, 22 September 2024 (UTC)[reply]
@Ymblanter: The Russian cultural heritage register has their own numbering system that has nothing do with the one from WLM Russia and there's already a property for it on Wikidata, P5381. So the database on WLM Russia's end serves no meaningful purposes what-so-ever outside of allowing them to control things related to Russian monuments. Regardless, there's no reason why they can't just drop the pointless intermediate numbering system and go with the exiting, official one from the Russian cultural heritage register. There's certainly nothing requiring them to use their own personal numbering system for monuments though and I'd argue it goes against the guidelines anyway. --Adamant1 (talk) 06:35, 22 September 2024 (UTC)[reply]
I am sorry but you do not know what you are talking about. The Russian cultural heritage register is so incomplete that it is almost useless. 90% of the monuments would have no number if we go with the official register. Ymblanter (talk) 06:47, 22 September 2024 (UTC)[reply]
Where did I say they were exact copies of each other or that the official list was as extensive as the one from WLM Russia? I could create my own numbering system for something right now that would include more things then an official list. That's besides the point though and has nothing to do with the merits of letting me use my own system on here. At the end of the day this is a media repository. That's it. It's not a database of monuments. Wikidata and Wikipedia exists for that purpose.
I'm not even saying they should get rid of the database. I'm just saying the template and the fact that they are using as a way to control things is inappropriate, totally pointless, and goes against the guidelines. I could give a crap if us not allowing for it causes them to take their ball and go home. That's not on us. They can always create a Wikidata property for it and add the information through infoboxes if the system is that useful. But it's totally pointless to have the numbering system as a template that gets added to specific categories. Things like that are exactly what Wikidata and infoboxes exist for. A lot of the monuments need to be documented better on Wikidata anyway. This is a perfect opportunity. --Adamant1 (talk) 07:04, 22 September 2024 (UTC)[reply]
I think turning the numbering system into a Wikidata property that can be added to infoboxes instead doing it through a template that's added to categories is perfectly clear. Your just being insincere about it. --Adamant1 (talk) 07:16, 22 September 2024 (UTC)[reply]
I uploaded this photographs, and it very hard to complain all unwritten rules for descript a cultural heritage. There is both category for building, category for code of building, some photographs lay in both categories, some in one of them. Then suddenly i learned that category for number - is not category, is looks like category, but not category, and i was not able to put it into subcategory. Later i found that i should manualy add name of newly created category of building in wikivoyage table, even table has wikidata id for building, and wikidata has commons category entered, it is not enough.
I am overwhelmed, and do not undersnand, how i should descripts photos. Why
I have rewritten the page in the light of the above discussion. However, I note that it includes a numbered bullet point, saying:
Categories by ID are automatically added by our templates and adopt simple names Category:WLM/ID and Category:WLE/ID for cultural heritage and natural sites, respectively. These categories are not created as pages, but added to all WLM/WLE images, which is sufficient for generating galleries of images with a given monument ID. Example here. Such galleries are linked from our lists of natural sites and cultural heritage under the name "галерея". These numbered categories are vital for sorting images and finding images with a given ID. If you are unhappy about red links generated by these categories, propose a better way of finding an image based on monument ID.
Unfortunately, this recipe is not implemented via the MediaWiki API, so the result will not be machine-readable. Destroying the WLM/xxx categories will break several tools on Toolforge and possibly some external applications. Olksolo (talk) 15:31, 23 September 2024 (UTC)[reply]
It is reasonable to allow some time for transition, so that such tools can be modified, before the categories are deleted. It would not be reasonable to use those to insist that the categories cannot be deleted or redirected. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits16:16, 23 September 2024 (UTC)[reply]
@Olksolo: I think the difference there is that the categories contain sub-categories for the ships and aircraft. Like if I go to Category:IMO 1000021 it contains Category:Montkaj (ship, 1995). Which is how it should be. The same can't be said here though. If you go to Category:WLM/2500622000 it just contains images that are also in Category:Pogranichnaya Street 2, Vladivostok. Which isn't how it should be. Also, what qualifies as a "cultural heritage monument" in Russia seems to be completely arbitrary and based on the personal opinions of WLM Russia members. Whereas IMO numbers are officially recognized and used by the International Maritime Organization among other organizations. Any number system we use on here should at least be semi-official and agreed on outside of small group of users though. I don't think we should allow for any user created numbering or categorization systems regardless of if it's by WLM Russia or anyone else. --Adamant1 (talk) 16:11, 23 September 2024 (UTC)[reply]
But you are mistaken.10-digit numbers of cultural heritage sites were officially used from 2004 to at least 2011. The "cultural heritage site passports" available in online sources often use 10-digit numbers (not the modern 15-digit ones).For example, look at the cultural heritage site passport on the website of the Ministry of Culture of the Russian Federation: [1] (documents from the Russian ministry may not be available if you are not in Russia, then a copy is [2]). At the very top of the page is not the modern 15-digit number, but the 10-digit one — 0300000170.For objects managed by the WLM Russia team, there are official documents on the cultural heritage status. Olksolo (talk) 16:37, 23 September 2024 (UTC)[reply]
I thought Ymblanter had said that the list from WLMs has more items in it then the official government one. If that's the case then they came up with the designation at least for the monuments that don't have official numbers associated with them. --Adamant1 (talk) 16:47, 23 September 2024 (UTC)[reply]
The problem is that not all cultural heritage sites with cultural heritage status are included in the official register. This work has been carried out by the ministry for several years with varying success. The peculiarity of bureaucracy in Russia is that this work will never be completed. But somehow it is necessary to identify cultural heritage sites with official status that are not included in the official register. WLM Russia Team has expanded the old 10-digit numbering scheme to identify such objects. Of the 223,374 cultural heritage sites supported by WLM Russia Team, approximately 45,000 are an extension of the old numbering. Olksolo (talk) 17:24, 23 September 2024 (UTC)[reply]
In an ideal world, when each cultural heritage site would have its own Wikidata Entity, this might be true. But now, out of more than 200,000 Russian cultural heritage sites, only about 70,000 have a Wikidata Entity. And the question of whether a Wikidata Entity should be created for all of these cultural heritage sites is still open. Olksolo (talk) 17:34, 23 September 2024 (UTC)[reply]
Any of them that have categories on Commons can and probably should have Wikidata items. If for no other reason then to track the artists and copyrights. --Adamant1 (talk) 17:48, 23 September 2024 (UTC)[reply]
There is a further issue. We have, for example a category Category:WLM/1010021052; note the Wiki Loves Monuments ID (P2186) used on Wikidata is RU-1010021052
(note the "RU-" prefix). Not only does this make them harder to match, but implies that there may be a "1010021052" monument in other countries' lists; so the Commons name is not guaranteed to be uniquely identifying.
And, if a Wikidata item is not found, or the Wikidata item is not linked to a category, or the Wikidata item is linked to a "WLM/1010021052" style category, write to a log file.
Hi, having seen some DRs recently related to copyright violations coming from Fandom, which has a CC-BY-SA disclaimer that only applies to text, I created a guide to re-using Fandom content here: Commons:Fandom files.
Please take a look, and go ahead and make any improvements. The primary goal is to try to clear up the misconception that images from Fandom are automatically CC-BY-SA.
This might be better as a subheading on Commons:Problematic sources; I don't think it needs to be a policy, since it's largely a summary of facts. Some details that I'd focus on are:
Some Fandom wikis use CC-BY-NC or -ND licenses for textual content, which are incompatible with Wikimedia licensing. (And don't ask me how a -ND license is supposed to work on a wiki.) Comments posted on Fandom wiki pages are not freely licensed at all.
Fandom does not take any steps to verify licenses on uploaded images. Even if a specific image on a Fandom wiki is specifically claimed to be freely licensed (which should appear under "more info" when viewing that image), users importing that content to Commons should take appropriate steps to ensure that license is accurate. Images which don't specifically show a license in the file view should be assumed to be non-free.
The description of Category:Photovoltaic power stations (on roofs) says "Photovoltaic solar panels (photovoltaic power stations) on roofs." -> Even if there is the mentioning of "power stations" in the description I think this category is meant to contain typical installations on residential buildings as well ("Solar cell panels on roofs" is a sub-category of it, for example). BUT Category:Photovoltaic power stations (on roofs) is a sub-category of Category:Photovoltaic power stations which is described as a large-scale photovoltaic system (which means big ground-mounted systems in most cases). So something has to change. P170 (talk) 12:27, 22 September 2024 (UTC)[reply]
I wouldn't consider roof solar panels to be Photovoltaic power stations. Most of the time whatever power they generate is passed down to a power station in the garage or something. --Adamant1 (talk) 12:31, 22 September 2024 (UTC)[reply]
To make a concrete suggestion: I am in favor of renaming Category:Photovoltaic power stations (on roofs) to Category:Rooftop photovoltaic systems and placing it in a new Category:Photovoltaic systems. The latter category should also include Category:Photovoltaic power stations. I have already consulted with the creator of Category:Photovoltaic power stations (on roofs), and he confirmed that he meant ordinary rooftop systems and supports my proposal. Similarly, the same approach should be taken with categories for facade PV, balcony PV, etc. --P170 (talk) 19:13, 23 September 2024 (UTC)[reply]
Correct. Per w:Power station, a power station is "an industrial facility for the generation of electric power". A "photovoltaic power station" is a large-scale facility whose primary purpose is generating solar power, not a set of solar panels which are incidentally installed on a building. Omphalographer (talk) 21:00, 23 September 2024 (UTC)[reply]
I'm aware of that and that's what the topic is about. Again my suggestion: 1 Photovoltaic systems 1.1 Photovoltaic power stations 1.2 Rooftop photovoltaic systems 1.2.1 Solar cell panels on roofs. If you omit 1.2 it would be inconsistent because 1.1 is about a whole system und your 1.2 (my 1.2.1) is not. P170 (talk) 15:59, 24 September 2024 (UTC)[reply]
How would the use of 1.2 and 1.2.1 differ? For instance, what would be an example of a photo which belongs in "rooftop photovoltaic system" but not "solar cell panels on roofs"? Omphalographer (talk) 19:16, 24 September 2024 (UTC)[reply]
A tram is successfully going there in the picture. The tracks don't look good but are we entitled here for a personal judegement if the pictured thing looks good enough? Andrei Romanenko (talk) 13:52, 28 September 2024 (UTC)[reply]
Hosting HDR images as JPEG with gain map
The tools for creating and displaying High Dynamic Range (HDR) images are starting to mature. HDR displays can render much brighter highlights than before, which leads to a big qualitative improvement in an image. Software for HDR production, and web-browser support, are becoming wide-spread. (Note that this is distinct from the tone-mapped HDR images you may have seen for the past decade or so.)
My test results are at Category:HDR gain-mapped images. Both images survived the process of uploading and rendering previews. HDR metadata was stripped from preview images, but preserved in the original uploads. If you have a newish HDR screen and a compliant web browser, the originals of this house and this church will appear brighter than usual. The effect on the house is subtle, limited to where sunlight hits white paint. The effect on the church is more dramatic: the windows should appear much brighter than the rest of the interior.
Most users of Commons images will see one of the smaller standard files, so for now the benefits of publishing this sort of content are limited. Are there any downsides to publishing it on Commons?
This post isn't marked as a proposal, because hosting these images on Commons works already. At a later date, when the standards are settled and the hardware is widely available, it would be nice to preserve HDR metadata in the generated preview images. — Preceding unsigned comment added by Semiautonomous (talk • contribs) 23:51, 23 September 2024 (UTC)[reply]
September 24
Are AI system capabilities of "Reading comprehension" higher than humans?
I very much doubt / disagree with that. I think this is another case of AI hype which doesn't seem to be a first for OWID.
The chart only says "Reading comprehension", and not for example "Measures of reading comprehension according to organization X" / test Y or anything of that sort. This is why I added {{Factual accuracy}} with the optional short explanation (there's lots of more sources for that) Testing for "Language understanding" and "Reading comprehension" is flawed and AI as of 2024 is not on or above human level for at least these. Many sources like this support that it's flawed. However Alenoach (talk·contribs) removed it arguing on the talk page GLUE and SQuAD are widely accepted as benchmarks for evaluating language understanding and reading comprehension. Even if that was the case, that chart does not say something like "according to GLUE and SQuAD measures" but is absolute so I think template would be warranted even if the user was right. Also see the other measures where "AI" is allegedly above human performance or approach it. Moreover, there's many tests that would suggest otherwise and these are not included or considered here so the accuracy is still disputed.
Moreover, I'd be interested if there is a place better suited to discuss accuracy issues like this or some project here otherwise engaged with accuracy issues with media on WMC such as working on identifying more of these. I don't know of a better place to ask and this could be closed and archived if solved or if a better place is found. --Prototyperspective (talk) 00:02, 24 September 2024 (UTC)[reply]
For information, here was my comment on the talk page:
I removed the "factual accuracy" because GLUE and SQuAD are widely accepted as benchmarks for evaluating language understanding and reading comprehension. These benchmarks are not perfect and have limitations, but they offer objective metrics to assess AI models in these areas. The linked article is quite opinionated, rejecting the idea that LLMs aim for truth and arguing that they merely "bullshit" (despite truthfulness being one of the goals of RLHF). And the article does not challenge the validity of GLUE or SQuAD as benchmarks for these specific tasks.
My impression is that the linked article is more of a philosophical essay that contends that LLMs really don't care about truth. This is a controversial claim, but even assuming this were true, are there major language understanding or reading comprehension benchmarks on which the latest frontier models have poor results compared to humans? On the other hand, I agree with you that it's a problem that the name of the tests isn't directly in the graph. Perhaps I could try to make a derived version based on the SVG, which would present it in a format like "SQuAD 1.1 (reading comprehension)" rather than "Reading comprehension", that would be more precise. Alenoach (talk) 00:57, 24 September 2024 (UTC)[reply]
The 2024 AI Index actually has a very similar graph, and doesn't mention Kiela et al.Similar image from the 2024 AI index It contains less information, but can be a good replacement to the image from Our World in Data if needed. Alenoach (talk) 18:05, 24 September 2024 (UTC)[reply]
Note that this is not about the particular source, there's many sources that support this (some examples: 1234567).
Even if some points of Alenoach were good and the recently added file description mitigated the problem a bit, the file accuracy template at least would still be due to how the chart is titled and subtitled. Prototyperspective (talk) 16:09, 24 September 2024 (UTC)[reply]
What the heck is "Reading comprehension with unanswerable questions"?? Is that like "Read Dostoevsky and tell me whether or not there is a purpose to human suffering?" Nosferattus (talk) 04:44, 26 September 2024 (UTC)[reply]
The contents are the equivalent of selfies, they do not meet any notability criteria, this is not facebook. What's interesting here is they were transferred from a Flickr account so not selfies as such.
Hi, I've created a category for this because this is a large set of images. This specific set seems important to me because depicts a girl in many activities, filling a gender gap in some of them (e.g. she is a maker, not many girls has this opportunity).
Also, if we're going to host such a large collection of images of this presumably non-public figure, it would be really nice to confirm that she is OK with this (via Fabrice), even if it is not strictly necessary legally. Otherwise, it just feels a bit... iffy from a privacy perspective. Two or three photos, fine, but 295? Nosferattus (talk) 04:36, 26 September 2024 (UTC)[reply]
Hmm, looks like Fabrice hasn't been active in the past 9 years. Does anyone else feel like there might be a privacy issue here? Aside from the huge number of photos, there's also lots of private information in the descriptions, like her age (14), birthday, names of her and her relatives. I'm not sure it violates any policies, but it definitely feels problematic. Nosferattus (talk) 05:06, 26 September 2024 (UTC)[reply]
They are already on his Flickr. Removing it here won't really change anything. From the context it's safe to assume he had consent to take the photos Trade (talk) 17:11, 26 September 2024 (UTC)[reply]
If they are of the same person, then they should be categorised under the name too. Regardless whether they are notable or not. In this particular case they are not notable. Broichmore (talk) 18:41, 27 September 2024 (UTC)[reply]
I tested this with a deletion request, and @Yann: deleted it, without reason. Why?
I also put in this deletion request, to be told by @RZuo: Not eligible for speedy deletion. If you disagree, nominate for deletion. Again why, delete the request? Yes, as it happens I disgree with this image being on commons, that's why I nominated it for deletion!
RZuo accused me of making a "rash accusation" in special:diff/929948378 in that DR. I accused him of violating policy. Furthermore, he did so prima facie. I am writing about it here because Yann already closed the DR, but that didn't stop RZuo from posting after that. — 🇺🇦Jeff G. ツ please ping or talk to me🇺🇦 16:35, 28 September 2024 (UTC)[reply]
Thanks Jeff, for putting me on the right track. I should have put in a Mass deletion request, rather than test the waters with a single image. Just not thinking straight. Also in the second case with the adding a category, was me being lazy and experimental.
A letter to the Haitian people of 25 december 1799.
News from Coni (Italy) arriving on 27 december, but the events are of course much earlier. There is an 27 december reaction in Paris. I would suggest to ignore this.
This is totally tangential but how do things like that work with the calendar that was being used during the French Revolution? Is it just ignored or do you translate it into the normal calendar or something? --Adamant1 (talk) 15:04, 24 September 2024 (UTC)[reply]
Interesting. I have some letters from the French revolution and slightly before then in France that I'm planning on scanning and uploading at some point. Hopefully next month if I have the time. Unfortunately I don't speak the language though. Any chance you'd be interested in helping me date them and translate the basics so they can be categorized properly and whatnot? --Adamant1 (talk) 15:54, 24 September 2024 (UTC)[reply]
It helps if you know some French, but it is not stricly necessary. Taking the text with OCR or taking over de text manualy, you dont need to know French. There is a smal problem: the spelling checker works with modern French, but gives errors for words spelled the ancient way. And there is a French verb conjugation dat is no longer used. The original spelling has to be respected. There are very few real errors and these can be treated with a correction template (the original text included errors has to remain accessible). However is the text is badly printed, it is sometimes difficult to read the letters. You wil get the feel of it. Examples A/M: bâtimens/bâtiments, déja/déjà, habitans/habitants; extinct Verbs: faisoient, formoient, commandoit (in modern French: past tense: commandait). But dont worry: 99 percent of the work is taken over the text verbatim. Later the text can be corrected and verified.Smiley.toerist (talk) 22:28, 24 September 2024 (UTC)[reply]
I worked the Journal de Bruxelles nr 100 1799 (78, 79).jpg and Journal de Bruxelles nr 100 1799 (80).jpg
How to make my file be selected as the media of the day on the main page?
Hello! I am one of those new users on Wikimedia Commons. I want that my file be selected as the media of the day on the main page. How do I do that? Please, help me. I'm pretty sure that the file I posted it is in the public domain and for the proof, there is it. — Preceding unsigned comment added by Heitor Gois (talk • contribs) 15:06, 24 September 2024 (UTC)[reply]
There's no French Wikipedia article. The few photos I've seen were not in use anywhere. He has a very limited number of hits on Google, mostly his own publications, videos. I found one interview--in a French regional newspaper. I'm willing to bet the majority have been uploaded to Commons to use as an image hosting site for whatever reason...wouldn't be my first choice! Bastique☎ appelez-moi!21:25, 24 September 2024 (UTC)[reply]
Yes, right. I deleted most book and album covers, for a start. Really too many portraits, and as these are not selfies, we need the photographer's permission. I also deleted the user page, and blocked him for a week. Yann (talk) 21:44, 24 September 2024 (UTC)[reply]
Mr. Boucheix does appear to be a singer with some published works - not that that's necessarily a difficult bar to meet nowadays - but notability by Wikipedia's standards seems unlikely to be met.
Certainly not. Images of text are rarely in scope on Commons, and these are no exception. If these were notable quotes, they'd belong on Wikiquote as text - but I seriously doubt that they are. The number of portraits seems entirely excessive as well.
Yes, there's definite copyright concerns here. The portrait photos like File:Bernard BOUCHEIX de REYVIALLES - Auvergne - mai 2024.jpg (of which there are a bunch) were clearly not taken by Mr. Boucheix himself, and the photographers aren't credited. The book covers and theater programs (which it looks like Yann just deleted) were clearly uncredited, potentially unlicensed derivative works as well.
@Gryllida I would definitely crop out the elements outside of the monitor. I don't know the relation between the main video and the CCTV video--but I'm not sure if cropping out the audio is necessary. Bastique☎ appelez-moi!00:00, 26 September 2024 (UTC)[reply]
I want to upload this image of Soviet chemist Tserevitinov, who died before 1947 and is clearly younger in the photo. I do not know anything about the author, but that is also the case with say Sergey Vavilov or Lenin, the former picture shot around the same time Tserevitinov died, and the author is unknown. What condition of the Russian public domain are they satisfying? Also can I use it to upload the image I proposed? Thanks, ExclusiveEditor (talk) 14:07, 25 September 2024 (UTC)[reply]
Thank you very much for asking here! At least the Vavilov image does give an original source. If the uploader had access to that source and could confirm that the image was published there anonymously, that would go a long way to establish its public domain status.
I think the issue with the image of Tserevitinov that you are proposing to upload is that we know absolutely nothing about its original publication.Felix QW (talk) 19:59, 25 September 2024 (UTC)[reply]
Just noticed that you mentioned you could rename the files as well. It's possible to rename the redirect without leaving a redirect behind, which functions almost the same as deleting it. ReneeWrites (talk) 16:35, 25 September 2024 (UTC)[reply]
September 26
User who creates useless categories
I found an user continue to create personal and useless categories even after warnings on the talk page. How should I deal with this? Should I report as vandalism? Bart Buchtfluß (talk) 03:06, 26 September 2024 (UTC)[reply]
@Adamant1: The problem is what Enhancing999 stated. Thanks for the comment.
@ReneeWrites: He certainly has a useful uploads, but he doesn't understand my advice on the talk page. If he continues this, I will report him to COM:AN/U. Thanks.
@Enhancing999: Certainly. Thanks. I warned once that "Category:Cross & Feather Architecture"(speedy deleted) is a personal and subjective theme and is not appropriate as category of Commons, though he created almost the same category, Category:Cross & Architecture again. He and I are both Japanese speakers, actually, but I couldn't understand what talking about. Perhaps he created Category:Cross & Architecture as a personal collection. All media in this category are attributed to him. Bart Buchtfluß (talk) 13:15, 26 September 2024 (UTC)[reply]
My guess looking at the category is that they were trying to create a category for images of crosses on buildings. As opposed to say ones at a gravesite or something. I could see where a category like that might be useful depending on the circumstances, but Category:Cross & Architecture clearly doesn't make sense and it seems like there's a broader consensus against similar "X subject and Y subject" type categories. So I don't know. Maybe it could be workable with the right name though. --Adamant1 (talk) 13:23, 26 September 2024 (UTC)[reply]
HELP! One user only throws out poetic description ideas of the category in this discussion and doesn't try to understand what categories are in Commons, nor does he listen to the advice of other users. He seems to think of categories on Commons as communal forums or SNS like Pinterest. I'm so sorry, but I cannot handle it anymore! Bart Buchtfluß (talk) 23:33, 27 September 2024 (UTC)[reply]
Panorama of Manhattan's West Side from Across the Hudson
Panorama of Manhattan's West Side from Across the Hudson
I was looking through Category:PD textlogo recently and there seems to be a lot of things that are licensed as "PD textlogo" that either cantain graphs at best or aren't logos to begin with at worst. To give an example, there's the various OneNote logos, which clearly contain graphics. there's also book covers, which clearly aren't logos even if made up of simple text. CD covers, which again aren't logos despite having simple text. despite having text. There's also weirdly some charts using the template. I think people get the point. It seems like the usage of "PD textlogo" is all over the place to the point of being essentially meaningless. So I'm wondering there should be more specific templates for simple works of art or things that contain text that aren't logos. Like a "PD-textlabel" for CD labels or a "PD-textchart" for simple charts that are mainly text. Maybe there's a better solution though. Anyone have any thoughts or ideas about how to deal with it? Adamant1 (talk) 13:18, 26 September 2024 (UTC)[reply]
It might be because the text hasn't been translated into other languages yet. It looks like the last update to other pages besides English was in May. That's the only thing I can think of though. --Adamant1 (talk) 16:04, 26 September 2024 (UTC)[reply]
Publishers info in newspapers
We have a name for the masthead on page one, but what is that paragraph on page two that contains info on the publisher called? I want to direct people to where the copyright symbol is/isn't in various newspapers. RAN (talk) 17:59, 26 September 2024 (UTC)[reply]
Maybe the credit line or word mark depending on the situation? My money is on it being the credit line. It's hard to say without an example though. --Adamant1 (talk) 18:31, 26 September 2024 (UTC)[reply]
Commons doesn't have one. The English-language Wikipedia does (en:MOS:ERA), which calls for consistency within articles but doesn't say one is preferred over the other, whichever one you choose depends on the context. ReneeWrites (talk) 14:59, 27 September 2024 (UTC)[reply]
Upcoming Wiki Loves Folklore in Bangladesh
We are excited to announce that, just like the previous edition, we are planning to organize Wiki Loves Folklore Bangladesh in 2025. As we prepare for this event, we welcome any recommendations or suggestions you may have. We’re eager to hear from photographers, organizers, and Wikimedians across Bangladesh to ensure the success of Wiki Loves Folklore 2025. The previous version of Wiki Loves Folklore Bangladesh was a booming success. We are hoping to continue the success in 2025 too. Your input will help us shape a better event!
Icarus005 (talk) 13:07, 27 September 2024 (UTC)[reply]
Hi all! I've been working on a page that lists all current photo/media contests that impact Wikimedia Commons. What I'm trying to do is create a quick easy and attractive navigation box that can be transcluded onto any page (and possibly the Main Pages) that lists all of these contests. I could use some help designing that navbox, seeing as template magic is not my forté.
Also, while I intend to keep that page current, I could probably use other people's support if they are interested. I get information primarily from announcements, such as the telegram groups for Wikimedia announcements and Wikimedia commons.
Hi, I've tried editing my watchlist several times to clean it up, because it's too big, but every time it produces a Timed out error. Is there any way to overcome the error? Or perhaps an administrator could completely clean up my list. Thank you for your collaboration--JotaCartas (talk) 23:14, 27 September 2024 (UTC)[reply]
thanks Graham, I've been trying for more than 15 minutes to edit raw watchlist, but it also gives a "page irresponsive" error. I've been an editor since 2009 (15 years with more than 1'500'000 editions)) and I've never cleared the watchlist, so it must be very large. Any help is welcome JotaCartas (talk) 23:50, 27 September 2024 (UTC)[reply]
OK, I stopped being lazy, and I'm going through the preferences, to see if I can find the solution. If nothing works, I'll ask for help here again, thank you any way JotaCartas (talk) 00:10, 28 September 2024 (UTC)[reply]
Hi, I just cleared de watch list. Path: Preferences -> Watchlist tab -> Clear your watchlist button. That was my solution. The Watchlist is now completely empty, but I had more than 400,000 entries and the number of warnings I was receiving was becoming overwhelming, regards JotaCartas (talk) 00:39, 28 September 2024 (UTC)[reply]
This is a new page on misinformation in Wikimedia projects, mainly about ways to mitigate it and practices it is currently being addressed on Commons. Secondarily, it also lists some Commons-relevant ways it Wikimedia projects may be used to mitigate misinformation and inaccuracies in the external world.
If you know of anything more to add, please do so.
I think many more files here need the {{Accuracy}} template, a dedicated place to discuss accuracy disputes was needed, a dedicated info page for readers confused about files with misinfo here was needed, the warning about Accuracy issues should also be displayed in the MediaViewer, and something that puts new version file overwrites under more scrutiny (like showing up in Wikipedia Watchlists) could be useful. --Prototyperspective (talk) 17:33, 28 September 2024 (UTC)[reply]
I'd love to see more stringent rules around user created "flags" and coats of arms. I'm not sure just adding the {{Accuracy}} template is adequate there in most cases, but then it's also not that easy to just have the images deleted either due to the gaps in the guidelines around the subject. Really that goes for other types of misinformation to. The project really lags behind in moderating and mitigating it compared to other sites. --Adamant1 (talk) 20:26, 28 September 2024 (UTC)[reply]
This text seems helpful to advise corporate accounts that try to add texts (almost any) at Commons. I used it the other day and read (or re-read) it.
The part about "Uploading of company logos, press images," could probably be improved. Currently it reads "in order to advertise or use them only in promotional material on other wikis is also prohibited, and such images will be deleted if they are found to violate our policies."
I want to upload a file, but I can't. It's saying that February 2, 2024 (content creation date) does not match the license (I read somewhere this is in the public domain in US). But the website (This one) is in US and was published on February 2 2024 (the article in the link). That doesn't make sense! — Preceding unsigned comment added by Susbush (talk • contribs) 14:09, 29 September 2024 (UTC)[reply]
Content which was published eight months ago is not in the public domain in the US (or anywhere) unless the author has explicitly placed it in the public domain. I see no indication that this would be the case for the images or text on the rss.com web page you linked. Omphalographer (talk) 20:27, 29 September 2024 (UTC)[reply]
September 30
Image not displaying at correct resolution?
I re-uploaded a better version of this image as the previous one was tiny and very low res:
It seems to have uploaded OK, but when I view it in the Media Viewer, or from one of the pages it's linked to, it seems to display the old image. Is that just me? Thanks for any help! --Gilgongo (talk) 18:05, 30 September 2024 (UTC)[reply]
did you try com:purge and clear your browsers' cache or change a browser?
Thanks - I didn't know about com:purge. That seems to have fixed it I think. And yes, I'll look at trying to make an SVG, although right now I've no idea how to. Gilgongo (talk) 21:36, 30 September 2024 (UTC)[reply]
This is clearly far from the Paris-Gare de Lyon. When I check for the first road crossing the rails overhead is Boulevard Poniatovsky see: permalink. And I see no other roadcrossing beyond this point (I checked up to Villeneuve Saint-Georges). Unless the photographer used a drone I am stumped.Smiley.toerist (talk) 20:29, 30 September 2024 (UTC)[reply]
Sorry, I did not notice the location. And the openrailwaymap is a bit misleading as it suggest that the road is underneath. However, this is an very busy urban motorway, where it is not posible to stop. Maybe the traffic was at a standstil.Smiley.toerist (talk) 20:42, 30 September 2024 (UTC)[reply]
Support The file name and caption are the first things you see when you select an image that's used on Wikipedia for better viewing, right before you click through to the file's own page. For most people it'll probably be the only information they'll see. This information is absolutely important enough that it should be mentioned in the file name. ReneeWrites (talk) 13:03, 11 September 2024 (UTC)
Oppose. Nice idea, but I prefer templates that can be translated and add properties. Latin letters in a filename are not a good clue in other scripts. We may have endless rename requests. File naming hacks are also not systematic; we do not routinely encode other properties in filenames. Glrx (talk) 15:47, 11 September 2024 (UTC)
.svg.png)
Comment from a filemover. If you want to apply this requirement to files after upload, you should amend Commons:File renaming to make it clear that lacking a statement of AI-generation in the filename is good cause for renaming. Either by adding a new numbered criterion or by finding a way to shoehorn it into an existing one (2 or 3, I'd guess). --bjh21 (talk) 21:01, 11 September 2024 (UTC)
Strong oppose no no no, file names shouldn't be used for that. This opens up Pandora's box for controversial subjects. This should be in the file description, not in the name. Multichill (talk) 12:53, 15 September 2024 (UTC)
Neutral I'm not sure this is the answer to dealing with the problems caused by AI images. I think another proposal for putting "AI generated" in file names of AI generated artwork would be a better solution. Since it's not so much where the images show up in search results, but that people often don't know the images are AI generated to begin with. People can still accidently get an AI generated image if they are pushed down in the search results. What we need is better ways to clearly identify and deal with AI generated artwork to begin with. I feel like it's mostly a curation issue at this point though. --Adamant1 (talk) 00:32, 20 September 2024 (UTC)
_showing_israel%27s_1948_and_1967_borders.svg.html)
.svg.png)
.jpg&action=history){kind=link}
.jpg&diff=next&oldid=928483698){kind=link}
