Topic: what, so are AI witch hunts acceptable behavior here now?

https://e621.net/takedowns/29101

so what, artists are now expected to "prove" they DIDN'T trace AI-generated images?

it always blows me away when people go "actually witch hunts are okay".

e_pluribus_eunuch said:
so what, artists are now expected to "prove" they DIDN'T trace AI-generated images?

Only when there's sufficient evidence to convince the mods that it AI was used. Notably it's not based on accusation or hearsay, but enough strong evidence that the mods can't ignore. They will not say what they consider to be good evidence so as to not give people wiggle room to work around it, and to remain flexible to change their methodology as AI changes in the future.

Nobody has said witch hunts are okay, staff only step in when they find sufficient evidence
There are 171 posts deleted for ai by the artist: deerwalker delreason:*ai*
I am sure there are plenty of discrepancies that could be pointed out in many of these posts

Also keep in mind that the site has multiple people that have been artists for years to decades, they know what they're looking for and what could be a human mistake and what might look like someone tracing something

a write-up of Spe's investigation are on the artist's wiki

On 2026-03-09, (...) spe told the artist that the only solid evidence she could provide would be a full screen recording of her art process. The artist agreed to provide a recording. However, on 2026-03-11, the artist left the Discord server without a word and cleared her e621 profile text

they agreed to a request for evidence then sat on it for two days and left without a word, that seems pretty damning

I find it funny that she says she was "given a requirement to provide something that is physically impossible" - all we asked for was a screen recording. Many real artists have been able to provide this in the past. That’s FAR from an impossible or even unreasonable request… unless, of course, she actually does trace AI, in which yeah, providing a recording that doesn’t show tracing would be impossible, lol.

Take a look at borkysleeve. Their art was flagged as AI the other day, but they provide full screen recordings in the description at the source. I unflagged it. Evidently, they know what their art looks like and decided to get ahead of the accusations, which was smart. But here we have an artist accomplishing this supposedly impossible feat for every individual post without even being asked. So yeah, I don’t buy it. Deerwalker is just trying to save face at this point.

That said, there is absolutely no malice or ill will on our part. If Deerwalker still wishes to prove her innocence, she is always welcome to submit that screen recording we asked for.

As someone who can look behind the scenes, I can tell you that these are no witch-hunts. The Janitors and future Blade Runners know they must produce very good evidence for their reports on AI pictures. Honestly, they've gotten pretty good at it. If they're seemingly being stubborn and refusing to "see reason", then there's a very good reason why.

Furthermore, take a moment to think about what an actual witch targeted by a "witch-hunt" is going to do, especially in an environment hostile to witches. Do you think they'll readily come clean and risk persecution? Or in order to survive, will they try to control the narrative, hide their witchiness, and try to pull the wool over the eyes of the inquisitors, even when they no longer have a leg to stand on?

"When a man is wrong and won't admit it, he always gets angry." ~ Thomas Chandler Haliburton, Canadian jurist and humorist

e_pluribus_eunuch said:
what, so are AI witch hunts acceptable behavior here now?

https://e621.net/takedowns/29101

so what, artists are now expected to "prove" they DIDN'T trace AI-generated images?

it always blows me away when people go "actually witch hunts are okay".

Just in case it isn't obvious already, you should absolutely NOT go on AI witch hunts yourself or make public accusations about suspected artists.
Only site staff are permitted to deal with AI claims, which is why there is also a new staff position being introduced specifically for that (i.e., Blade Runners).

All accusations are dealt with privately through the post flagging system, and with enough evidence and a thorough investigation, the admins would determine whether or not an artist's work should be purged due to suspected AI use.
All the artist has to do to counter against these claims is to show some concrete evidence that their works are genuine and human generated, which the artist in question had supposedly failed to do so.

clawstripe said:
As someone who can look behind the scenes, I can tell you that these are no witch-hunts. The Janitors and future Blade Runners know they must produce very good evidence for their reports on AI pictures. Honestly, they've gotten pretty good at it. If they're seemingly being stubborn and refusing to "see reason", then there's a very good reason why.

Could you perhaps shed some light into what happened with Oddscreen here? This appeared, at least from the outside, a situation that was poorly handled. Is it still the practice to nuke all content and label it as AI before attempts to reach out to the artist in question for proof? Or if you're still doing that, is there anything in place to ensure prompt response to when the artist does reach out? Because with accusations that have such negative stigma as being an AI artist an excuse of "we're busy" is rather underwhelming for why it took so long to get back to them, and I can't help but feel Oddscreen would've had to wait even longer if not for the pressure on the forums.

Again I understand I'm only seeing things from the outside, so there could be plenty of missing context, but if what was seen from the outside was largely how things played out I hope procedures have improved since then. The case mentioned in this thread would seem to suggest things are better, so perhaps it's a non-issue that I'm bringing up.

sylenial said:
-snip-

donovan_dmc said:
The accuser presumably provided some proof, then the staff did their own investigating
There is not a world in which no investigation happened and the artist was just tossed out the window

It also isn't going to "become a bigger problem", the staff here don't take ai accusations at face value
I literally was staff a few months ago, I have been part of ai investigations, multiple people are involved and everyone points out things and attempt to come up with plausible explanations
Deletions only go forward if multiple people agree there is no plausible explanation for some issue(s)
And even then the artist is only nuked entirely if a pattern of issues is found, else just the post(s) with issues would be deleted

Also, the situation you're referring to has been dealt with way over a year ago, and their posts seemingly have been reinstated. So, it seems that, as stated earlier with a different user, evidence was provided, and their posts were recovered.

i know this feeling, i just got posting privileges banned as well.
got told its cause i used traces and such 2 years ago to make a pony piece and hadnt posted anything since, so had no art history. asked if all new artist got treat this way and i quote "no only you"

provided photos of my sketch pad, showing the original drawings, told them all i did was scanned them, deleted the paper back ground and in some cases, would cut parts and move them or warp them. before putting them togther ro do a page of faces.

but because i cant provide a file showing the faces in the layers while i was cleaning up. apparantly i have no evidence and im forging evidence.
how does one forge a photo?

As a note related to the above post, I have already prepared an explanatory writeup for the above artist concerning them misrepresenting my complaints, threatening me with screenshots of our Discord conversation, taking photos of sketches with significant visual differences from the AI-generated versions they uploaded at post #6265502, and more.

The full writeup will be at the ravenfalls (artist) wiki article after the co-investigators in this case have had the chance to review my work.

My conscience is clear and I simply will not be manipulated by attempts to threaten me or sway public opinion. The evidence speaks for itself.

Had this conversation before. The admin know. The rest of us are expected to believe.

bird-tm said:
Also, the situation you're referring to has been dealt with way over a year ago, and their posts seemingly have been reinstated. So, it seems that, as stated earlier with a different user, evidence was provided, and their posts were recovered.

The issue I have is that their art was nuked with the assertion it was AI before any apparent attempt to reach out to the artist for them to prove otherwise, staff didn't get back to the artist promptly when attempts to prove otherwise were made until nearly a week later (and possibly after a bunch of pressure on the forums), and ultimately it was clearly determined that the art was removed in error.

My concern is that attempts to reach the artist should have been done before removing all their art and labeling the artist as an AI artist because the accusation is so damning and has so much stigma attached to it. And barring that at the very least the artist is owed prompt attention to their attempts to prove their innocence, which this artist didn't seem to get.

I hate AI artists getting away with pretending to be a real artist, but I hate real artists being prematurely labeled as AI artists in a "guilty until proven innocent" approach even more. The onus should be entirely on E621 staff to show they made a good faith effort to clear things up with the artist before removing their art.

Once again this is simply how it appears from the outside and I only have one perspective to go off of which is why I was hoping to get confirmation that either I am in fact missing important context that would change the perceived situation (and I don't even care to know what the context is, just that there is missing context), or that changes have taken place to improve the whole process for all parties.

grizzly_yote said:
Had this conversation before. The admin know. The rest of us are expected to believe.

lafcadio said:
As a note related to the above post, I have already prepared an explanatory writeup for the above artist [...]
The full writeup will be at the ravenfalls (artist) wiki article after the co-investigators in this case have had the chance to review my work.

Be constructive with your criticism; if this isn't enough, then what is? If you want us to get better, you could at least try to explain why the status quo is insufficient.

We provide write-ups in many cases, & people seemingly refuse to read them or are simply unaware of their existence. You were told that this one is coming; what's your justification for not evaluating the evidence when it's provided & making your own call?

aacafah said:
We provide write-ups in many cases, & people seemingly refuse to read them or are simply unaware of their existence. You were told that this one is coming; what's your justification for not evaluating the evidence when it's provided & making your own call?

When some or even all of their favourite artist's posts are gone, if I were them, I would check comment sections first and then forums... I guess few people would notice the change on the artist wiki page.
I'm not sure when e6 started to write the AI investigations there. Was there any announcement for this?

As for the content itself, investigations seem to be detailed and valid. They even provide the date and who was involved.

yummytummysketties said:
When some or even all of their favourite artist's posts are gone, if I were them, I would check comment sections first and then forums... I guess few people would notice the change on the artist wiki page.
I'm not sure when e6 started to write the AI investigations on the wiki page. Was there any announcement for this?

As for the content itself, investigations seem to be detailed and valid. They even provide the date and who was involved.

There was no announcement, but we did sort this out in a staff channel. I intend to draft some more formal pages now that we're recruiting for a more formal "Bladerunner" team, but I can explain this here.

2025-09-28, I brought up the topic of people editing artist wikis to go "AI, do not post". Letting people just do this completely unchecked seemed like a bad idea, and I also floated the idea of us doing this ourselves to as further deterrence against wiki vigilantism. Anybody could fake something using our template, but only staff could actually lock the affected wikis. A mere 10 days ago I had concluded a case against an artist. This wasn't something I would actually deploy for public writeups against specific artists until a few days later, 2025-10-01.

aacafah said:
Be constructive with your criticsm...

Okay.

Every time I've said anything to admin in these forums or by private message, the responses have been extremely patronizing.

There's no humility and no willingness to even consider you might be wrong.

Nor do you give any evidence that I've seen. The best I've found is "our expert staff have extremely good reason to believe." And we're just expected to accept that. (Where does this exist?)

It's frustrating and off-putting.

is there a defined N 'bladerunners' youre trying to get? like a few AI-vestigators increases the odds of some goofy false positive drama if they're the sole adjudicators. more seems like it could lead to artist-janitor interpersonal drama.

I remember when this was the way people treated every image when Photoshop was first a thing.

People learn nothing, And Photoshop is still around.

pocket_erector said:
I remember when this was the way people treated every image when Photoshop was first a thing.

People learn nothing, And Photoshop is still around.

if photoshop makes an inference of what some other artist might draw and generates a sample of it, than ig they're the same. otherwise not rlly the same situation

oneohthrix said:
if photoshop makes an inference of what some other artist might draw and generates a sample of it, than ig they're the same. otherwise not rlly

Tools are tools, And uneducated panic is uneducated panic.

pocket_erector said:
Tools are tools, And uneducated panic is uneducated panic.

I mean, that's kind of a really reductive view.

everything is a tool. that's not, like, a qualitative statement. items being within the same set does not make them not dissimilar.

dba_afish said:
I mean, that's kind of a really reductive view.

everything is a tool. that's not, like, a qualitative statement. items being within the same set does not make them not dissimilar.

a strong statement like 'X is X, Y is Y' prob needs fact checking imo. might be misinformation

pocket_erector said:
Tools are tools, And uneducated panic is uneducated panic.

"Tools are tools" is an interesting way to put it when Photoshop is different from AI image generation in every conceivable way.

Photoshop uses very basic math to do very basic editing to multiple pixels and it actually takes significant artistic effort to get a meaningful result. Everybody knows how the Multiply blend mode works, it multiplies two pixels' color together.

AI image generation uses unknowable devil magic (a neural network with so many moving parts as to make its inner mechanisms completely impossible for a human to understand) to conjure entire images from nothing but stolen Danbooru data. Nobody knows how to flip the bits that make the AI better/worse at copying Salvador Dali.

Which is to say, they're not comparable at all and the clueless AI bros are the only ones who even try to equate them.

grizzly_yote said:
Okay.

Every time I've said anything to admin in these forums or by private message, the responses have been extremely patronizing.

There's no humility and no willingness to even consider you might be wrong.

Nor do you give any evidence that I've seen. The best I've found is "our expert staff have extremely good reason to believe." And we're just expected to accept that. (Where does this exist?)

It's frustrating and off-putting.

I am not aware of any appeals by you, but we do try to conduct appeals by leaving a clear path towards an innocent verdict because the nature of AI assistance is just that insidious. All the data comes from humans, so nearly everything comes down to something an artist might have actually done in the past.

The biggest problem remains that AI artists have the exact same reading ability as average e621 users, so anything we divulge is something we should expect to eventually find its way to AI artists. If I post our complete training docs publicly, any reasonably aware AI artist suddenly has a recipe to avoid detection basically forever.

Obviously being a completely unaccountable and secretive investigation team would not do much to inspire users' faith, but the more we let on, the more ineffective this entire practice becomes. Providing occasional writeups for cases where the artist admits to it or can never appeal again (for instance, the above case where I have shared definitive proof with other staff that this appellant was willing to forge evidence), and occasionally stepping in to prevent vigilantes from causing a fuss, seems like the only realistic option.

I would be interested to know if there was some alternative method of investigation/disclosure that would perfectly convey everything to people who don't use AI, while simultaneously teaching nothing to those who do.

lafcadio said:
As a note related to the above post, I have already prepared an explanatory writeup for the above artist concerning them misrepresenting my complaints, threatening me with screenshots of our Discord conversation, taking photos of sketches with significant visual differences from the AI-generated versions they uploaded at post #6265502, and more.

The full writeup will be at the ravenfalls (artist) wiki article after the co-investigators in this case have had the chance to review my work.

My conscience is clear and I simply will not be manipulated by attempts to threaten me or sway public opinion. The evidence speaks for itself.

i didnt threaten you, show a single part in the chat where i threatened you. at all?
yes ive taken screenshots and have them all saved so show your bully tactics.
and as one of the guys i work with has pointed out, did i ask you to look at the meta data for the image, as i cant change that. oh wait that all shows was taken from a camera, from a phone.

but hey, you were fine with the photos till i didnt have a saved version of my scans.

lafcadio said:
There was no announcement, but we did sort this out in a staff channel. I intend to draft some more formal pages now that we're recruiting for a more formal "Bladerunner" team, but I can explain this here.

2025-09-28, I brought up the topic of people editing artist wikis to go "AI, do not post". Letting people just do this completely unchecked seemed like a bad idea, and I also floated the idea of us doing this ourselves to as further deterrence against wiki vigilantism. Anybody could fake something using our template, but only staff could actually lock the affected wikis. A mere 10 days ago I had concluded a case against an artist. This wasn't something I would actually deploy for public writeups against specific artists until a few days later, 2025-10-01.

Any chance of you guys mentioning the wiki entry edit in deletion reasons? It'll be easier for us to find.

ravenfalls said:
i didnt threaten you, show a single part in the chat where i threatened you. at all?
yes ive taken screenshots and have them all saved so show your bully tactics.
and as one of the guys i work with has pointed out, did i ask you to look at the meta data for the image, as i cant change that. oh wait that all shows was taken from a camera, from a phone.

but hey, you were fine with the photos till i didnt have a saved version of my scans.

I am allowed to read between the lines when you suddenly make mention of screenshots after challenging me to prove your plagiarism by posting the original images that you have. As a show of good faith, I will preempt you by posting the logs in their entirety.

I don't recall ever saying I was fine with your photos. It was very much the opposite, as within 3 minutes of downloading the evidence I already had findings to share with other staff. One of your edits was to add a mistake to a perfectly straight line. Even the moderator who handled your latest record has seen it. Along with the janitor and admin who saw your work before I did, that makes four of us. Your fraud is understood by multiple staff members besides myself.

Your article is now live at ravenfalls (artist). I will give you no further replies.

yummytummysketties said:
Any chance of you guys mentioning the wiki entry edit in deletion reasons? It'll be easier for us to find.

There's a bit of a technical hurdle I see here (what if the deletions come before the wiki does) but nevertheless I've left this as something for interested coders to potentially tackle.

lafcadio said:
I am allowed to read between the lines when you suddenly make mention of screenshots after challenging me to prove your plagiarism by posting the original images that you have. As a show of good faith, I will preempt you by posting the logs in their entirety.

Reason for contact:
i have had 2 hand draw images with some photoshop clean up deleted being told that its ai?

Tickets will automatically close after 5 days of inactivity.

Lafcadio — Yesterday at 2:39 PM
What are you willing to share in your defense?
RavenFalls — Yesterday at 2:50 PM
what do you want me to share? i hand drew some images, put them on photoshop so they could all be on one screen, cleaned them up and uploaded the.
some of the faces still had the guide lines on them.
Lafcadio — Yesterday at 2:53 PM
Is there any proof that you, specifically, drew them? Do you have layered source files to show us? A video process? Anything that might inspire confidence in your artistic ability?
Mind, the last thing we have on record from you is a series of 9 MLP posts that look like this.
https://e621.net/posts/4716550
https://e621.net/posts/4706772
Lafcadio — Yesterday at 2:55 PM
I'll note that if you need to make a video for us I would like to provide very specific directions for it. If you already have a video or can't submit one then we won't worry about that for now.
RavenFalls — Yesterday at 2:58 PM
im not refering to those two images, those pony images were fairly removed for me using backgrounds from the show
6265502 and 6259753
Lafcadio — Yesterday at 3:00 PM
Actually, I'm the one who deleted these MLP posts, and the use of backgrounds is the least of your concerns. I bring them up now because, from my perspective, you have moved from one method of editing/plagiarism to another.
RavenFalls — Yesterday at 3:02 PM
but im not refering to them though, im refering to the resent concept art, for the game ringbound, the project i am currently working on.
and if you suspect me of plagiarism, where have i got the images from?
Lafcadio — Yesterday at 3:06 PM
To briefly recap:

• You post these MLP collages.
• Almost two years pass.
• You post these fox drawings.

If you were merely posting AI images, then the obvious conclusion is that you've had no art to share over the last two years. If there was actual, significant artistic history, then maybe you'd be able to share evidence of it.
Lafcadio — Yesterday at 3:09 PM
[ and if you suspect me of plagiarism, where have i got the images from? ]
Unless you also want to contest the deletion of the MLP images then I will just end this particular line of inquiry with "this isn't how it works".
RavenFalls — Yesterday at 3:16 PM
no, at no point have i said i want to contest the pony deletion, have i even mentioned them, no. you brought them up.
those were edits we used as part of a dnd campaign many years ago. i have no interest in them.

the only images i have mentioned are the two recent ones, the ones for a game i am working. why on earth would i have continued to post art work between then and now? when obviously the edits broke the rules and i accepted that.
and i only started working on ring bound about a month ago?
Lafcadio — Yesterday at 3:19 PM
If you cannot prove any prior art history and cannot prove the creation of these pieces then there is no basis for you to contest my findings. I am not currently accepting these kinds of arguments. Post hard evidence or escalate to higher staff (I will share my findings with higher staff to show to them exactly, why, we consider your work to be suspicious.)
RavenFalls — Yesterday at 3:21 PM
so what your saying is, every new artist, who has not posted art work before, is stealing art, thats a great policy
Lafcadio — Yesterday at 3:21 PM
Not every new artist, just you.
RavenFalls — Yesterday at 3:22 PM
so im being discriminated against, for posting a new image. so essentially i am banned from being part of the site
Lafcadio — Yesterday at 3:27 PM
As your account has previous deletions from using other people's MLP art, if you can't prove human creation of new work that has obvious traits found in AI-generated art, then I do have cause to delete individual uploads, and disable your uploads entirely if needed. Uploading somebody else's art with permission would be another story, but the evidence so far is not in your favor.
RavenFalls — Yesterday at 3:30 PM
what evidence of ai? theres isnt any evidence. you have openly admited and i have a screen shot of it stating my work was taken down due to work i subbited 2 years ago. you removed my current hand drawn work, because you have the power to do so even though ive done nothing wrong.
Lafcadio — Yesterday at 3:34 PM
I am under absolutely no obligation to share my evidence with you. We withhold this kind of information because guilty artists can use our evidence to falsify less suspicious art later on.

You are additionally misrepresenting my argument by saying that this solely comes down to your old MLP art. Even the new art is visibly suspicious, as evidenced by the fact that your art was brought to my attention by another janitor.

This style of dialog suggests to me that you are willing to try and fight me socially without ever submitting evidence. My conscience is clear and the evidence is readily available for other staff members, so I will simply not allow this kind of blustering to happen.

Either submit evidence, escalate this to higher staff, or close this ticket to end the appeal. I am not leaving you any more options.
RavenFalls — Yesterday at 3:37 PM
[ To briefly recap:

• You post these MLP collages.
• Almost two years pass.
• You post these fox drawings.
• If you were merely posting AI images, then the obvious conclusion is that you've had no art to share over the last two years. If there was actual, significant artistic history, then maybe you'd be able to share evidence of it. ]

you have clearly stated here, because i have no history from two years ago to my current project is why its removed.
theres also nothing ai im my images, why on earth would ai use guide lines.
but you want proof, i will take photos of my sketch book when im home.

at which point i best get an apology
Lafcadio — Yesterday at 3:38 PM
I will await your submissions.
RavenFalls — Yesterday at 3:39 PM
you have also stated "If you cannot prove any prior art history and cannot prove the creation of these pieces"
once again stating in there, because i have no prior art history.
that your only targeting me
Lafcadio — Yesterday at 3:40 PM
You're blustering again. I will give your account a 10-minute timeout for now, as you clearly won't be able to provide any evidence until you're home. Please use this time to carefully consider how you will approach the remainder of this appeal.
As I have mentioned previously, we will not disclose the specific evidence used to justify these two recent deletions. This is standard procedure. The appeal process has always worked like this. It would be folly for us to divulge specific evidence simply because an appellant said "there is no AI".
RavenFalls — Yesterday at 8:31 PM
[ evidence.rar ]
while im at it, heres the photoshop line work for the next one, when you decide you want to delete that as well
[ next.psd ]
RavenFalls — Yesterday at 8:39 PM
and so its stated, shes ment to have an extra finger, that was even in the tags
Lafcadio — Yesterday at 9:48 PM
These traditional versions seem to have some significant differences from the digital versions you uploaded. If you did cleanup, do you have pre-cleanup digital versions to share?
RavenFalls — Yesterday at 10:25 PM
The preclean up versions are just scans or the pages.
I scan up, do a little warping and fix my mistakes and cut them onto a single page.
Of the pages*
Lafcadio — Yesterday at 10:42 PM
So you didn't do any cleanup separately from scanning/digitization?
RavenFalls — 4:57 AM
As i said.
I scan them.
I load them into photoshop.
I will erase any kind of major mistakes.
I might do a little warping if needed.
Theres a couple i cut the ear and repasted them.
They have the background paper deleted and moved to a blank sheet.

I dont need to do major editing, the drawing are pretty good to start.
Lafcadio — 8:17 AM
If things are as you described, it is completely impossible for your digital versions to be based on the physical versions you have submitted. You have already established that you do not have any pre-cleanup versions to share to bridge the gap, so the only reasonable explanation is that the differences I am seeing come down to you getting sloppy when trying to make traditional recreations.

As you have shown a willingness to forge evidence, your pre-AI work has its own signs of not being entirely honest, and the new work also triggers several other heuristics, I will be disabling your uploads from this account and making a writeup for public consumption that explains the nature of the deception.

I don't recall ever saying I was fine with your photos. It was very much the opposite, as within 3 minutes of downloading the evidence I already had findings to share with other staff. One of your edits was to add a mistake to a perfectly straight line. Even the moderator who handled your latest record has seen it. Along with the janitor and admin who saw your work before I did, that makes four of us. Your fraud is understood by multiple staff members besides myself.

Your article is now live at ravenfalls (artist). I will give you no further replies.

There's a bit of a technical hurdle I see here (what if the deletions come before the wiki does) but nevertheless I've left this as something for interested coders to potentially tackle.

ive read the write up, yes im going to take screen shots of our chat, as stated the mods are known for bully tactics, i am going to take screenshots so you cant try and change things and make a new narrative down the line.
like how the write up says i never gave you raw images, even though you was give photos from a phone camera. that by its own internal date stamping has the last change stamp as the same time as the photo taken, something i cant change.

next you go on about, how i have gave you my next piece that shows all all layer lines. stating that thats some how all ai gen, and taken from a final piece. if its from a final piece, why would i need the layer lines? or even upload that for approval before you kick off at my next image, because apparently having an extra finger for a plot hook is a bad thing?

to this point i would like a single artist on this entire site, that draws by hand, scans the image into their computer, then saves, then erases a few lines then save a new image, then maybe does an extra touch up, then saves another image.
why would anyone do that?

then you go on about how i photoshopped my evidence as it was different in places to my uploads, if i photoshopped then, they would be identical. but once again, since the data on the photos shows between the photo being taken and the last change being made being 0 seconds, i must have amazing photoshop skills.

i also love the last part, where you bring up the dating of the photos, and how they were taken after the appeal started, yes, because why on earth would i take photos of a sketchpad before then? and once again. why would i save a scanned image?

end of day, artists on here just need to be popular with mods. when a mod decides they dont like your work or the person. they just arnt allowed to do anything.

ravenfalls said:
ive read the write up, yes im going to take screen shots of our chat, as stated the mods are known for bully tactics, i am going to take screenshots so you cant try and change things and make a new narrative down the line.
like how the write up says i never gave you raw images, even though you was give photos from a phone camera. that by its own internal date stamping has the last change stamp as the same time as the photo taken, something i cant change.

next you go on about, how i have gave you my next piece that shows all all layer lines. stating that thats some how all ai gen, and taken from a final piece. if its from a final piece, why would i need the layer lines? or even upload that for approval before you kick off at my next image, because apparently having an extra finger for a plot hook is a bad thing?

to this point i would like a single artist on this entire site, that draws by hand, scans the image into their computer, then saves, then erases a few lines then save a new image, then maybe does an extra touch up, then saves another image.
why would anyone do that?

then you go on about how i photoshopped my evidence as it was different in places to my uploads, if i photoshopped then, they would be identical. but once again, since the data on the photos shows between the photo being taken and the last change being made being 0 seconds, i must have amazing photoshop skills.

i also love the last part, where you bring up the dating of the photos, and how they were taken after the appeal started, yes, because why on earth would i take photos of a sketchpad before then? and once again. why would i save a scanned image?

end of day, artists on here just need to be popular with mods. when a mod decides they dont like your work or the person. they just arnt allowed to do anything.

While I understand your frustration, this is not something you should bring up in forums. I'm not sure if you still have an open ticket on discord but please write those things there.

And it's not about popularity. You have a history of similar posts. That's why they're suspecting you.

ravenfalls said:
end of day, artists on here just need to be popular with mods. when a mod decides they dont like your work or the person. they just arnt allowed to do anything.

Just so you know, a layered work file can absolutely be faked by AI as well, which is why the staff investigating the matter would have to examine such evidence with extra scrutiny and increased doubt when there are "unconventional" drawing practices.

ravenfalls said:
ive read the write up, yes im going to take screen shots of our chat, as stated the mods are known for bully tactics, i am going to take screenshots so you cant try and change things and make a new narrative down the line.

Honestly this is not as personal as it feels. Consider the sheer volume of business a janitor has to deal with in a day. People upload AI generated or assisted art all the time, every day of the week... if janitors/blade runners took the time to bully everyone who gets investigated they would never get anything done.

There is nothing about you or your artwork that makes you so special that any janitor would feel the need to single you out and bully you. What happened to you is standard procedure for any artist who posts AI generated/assisted images. If you felt backed into a corner or that any evidence you could provide wasn't enough there was a reason for that.

Wasn't there someone who quoted a saying "A man who is wrong but won't admit it will always yell"?

lafcadio said:
I am not aware of any appeals by you, but we do try to conduct appeals by leaving a clear path towards an innocent verdict because the nature of AI assistance is just that insidious. All the data comes from humans, so nearly everything comes down to something an artist might have actually done in the past.

The biggest problem remains that AI artists have the exact same reading ability as average e621 users, so anything we divulge is something we should expect to eventually find its way to AI artists. If I post our complete training docs publicly, any reasonably aware AI artist suddenly has a recipe to avoid detection basically forever.

Obviously being a completely unaccountable and secretive investigation team would not do much to inspire users' faith, but the more we let on, the more ineffective this entire practice becomes. Providing occasional writeups for cases where the artist admits to it or can never appeal again (for instance, the above case where I have shared definitive proof with other staff that this appellant was willing to forge evidence), and occasionally stepping in to prevent vigilantes from causing a fuss, seems like the only realistic option.

I would be interested to know if there was some alternative method of investigation/disclosure that would perfectly convey everything to people who don't use AI, while simultaneously teaching nothing to those who do.

First, thank you. I genuinely appreciate the tone of this message. I feel like you're speaking with me and not at me, and for me that's important.

I've made no appeals because I wasn't allowed to. I commissioned artwork based on the recommendations of friends. I am adamantly anti-AI. People I trust reassured me the art was human drawn. I spoke with the artist and saw process images while he worked on it. The artist uploaded it here, and admin I don't know and have no relationship with took it down as AI. When I protested, I was told the artist has to appeal.

In the mean time, every conversation I've had with admin here has just felt demeaning. They've involved logical fallacies and condescending tones. And that's left me not even wanting to speak with them. I realize that comes through in my tone now, especially since I'm Gen X and speak fluent sarcasm once I reach a certain level of salt. In this case, that's how I got here, so again, I appreciate your openness.

grizzly_yote said:
First, thank you. I genuinely appreciate the tone of this message. I feel like you're speaking with me and not at me, and for me that's important.

I've made no appeals because I wasn't allowed to. I commissioned artwork based on the recommendations of friends. I am adamantly anti-AI. People I trust reassured me the art was human drawn. I spoke with the artist and saw process images while he worked on it. The artist uploaded it here, and admin I don't know and have no relationship with took it down as AI. When I protested, I was told the artist has to appeal.

In the mean time, every conversation I've had with admin here has just felt demeaning. They've involved logical fallacies and condescending tones. And that's left me not even wanting to speak with them. I realize that comes through in my tone now, especially since I'm Gen X and speak fluent sarcasm once I reach a certain level of salt. In this case, that's how I got here, so again, I appreciate your openness.

I mean you need to be careful with some artists these days. I had been duped several times. Especially by one who I thought was the real deal as I swore I heard her name through the grapevine. I have since learned my lesson

colinthelucario said:
I mean you need to be careful with some artists these days. I had been duped several times. Especially by one who I thought was the real deal as I swore I heard her name through the grapevine. I have since learned my lesson

Yes, obviously. That's why I rely on vetting information from friends I trust, as I said. I did my due diligence before spending money on art.

grizzly_yote said:
First, thank you. I genuinely appreciate the tone of this message. I feel like you're speaking with me and not at me, and for me that's important.

I've made no appeals because I wasn't allowed to. I commissioned artwork based on the recommendations of friends. I am adamantly anti-AI. People I trust reassured me the art was human drawn. I spoke with the artist and saw process images while he worked on it. The artist uploaded it here, and admin I don't know and have no relationship with took it down as AI. When I protested, I was told the artist has to appeal.

In the mean time, every conversation I've had with admin here has just felt demeaning. They've involved logical fallacies and condescending tones. And that's left me not even wanting to speak with them. I realize that comes through in my tone now, especially since I'm Gen X and speak fluent sarcasm once I reach a certain level of salt. In this case, that's how I got here, so again, I appreciate your openness.

It is true that artists are better at appealing because we can expect them to have some proof a commissioner wouldn't, but there nevertheless exists the possibility that a commissioner closely associated with an artist would have the kind of layered source files or high-detail recordings we normally accept. If you want to let me know which artist/post you were attempting to appeal I can try and give a second opinion.

lafcadio said:
It is true that artists are better at appealing because we can expect them to have some proof a commissioner wouldn't, but there nevertheless exists the possibility that a commissioner closely associated with an artist would have the kind of layered source files or high-detail recordings we normally accept. If you want to let me know which artist/post you were attempting to appeal I can try and give a second opinion.

The artist I commissioned was Hyena Beans. And I only reveal that so you don't think I'm being vague or slippery in light of what I say next. If I can be painfully honest, the validation isn't worth the struggle at this point.

When I look at other cases, it's plain to me the reviewers have made up their minds and can't be convinced otherwise. Artists present their cases, show their evidence, and the reviewers stand their ground. It always brings up more questions or isn't sufficient to sway them in their expertise.

Even if I did convince you, what would I get? A piece of art I paid for would be "archived" on e621, so that my intellectual property can help Bad Dragon profit off ads that harvest data from everyone who sees it. That kind of Google surveillance is almost as evil as AI is.

So, it's fine if you don't review it. I think the outcome is already a foregone conclusion. Even if you approved it, I don't want my IP to contribute to the surveillance and data aggregation economy.

grizzly_yote said:
Bad Dragon profit off ads that harvest data from everyone who sees it. That kind of Google surveillance is almost as evil as AI is

The ads here have nothing to do with google, they are done entirely in house

donovan_dmc said:
The ads here have nothing to do with google, they are done entirely in house

Kudos! Sincerely. I'm glad to hear that.

You know, the reason I made the bladerunner role is because these AI investigations are very time consuming. We don't just eyeball it and say 'eeeyup, thats AI because it looks like it'. We spend hours and hours going through their history, comparing art, searching for details and inconsistencies, then we work with them one on one to go over sketches they provide or screen recordings. The process usually takes several days and many hours to document and come to a consensus.

It might be imperfect, but it is very much not a witch hunt.

grizzly_yote said:
The artist I commissioned was Hyena Beans. And I only reveal that so you don't think I'm being vague or slippery in light of what I say next. If I can be painfully honest, the validation isn't worth the struggle at this point.

When I look at other cases, it's plain to me the reviewers have made up their minds and can't be convinced otherwise. Artists present their cases, show their evidence, and the reviewers stand their ground. It always brings up more questions or isn't sufficient to sway them in their expertise.

Even if I did convince you, what would I get? A piece of art I paid for would be "archived" on e621, so that my intellectual property can help Bad Dragon profit off ads that harvest data from everyone who sees it. That kind of Google surveillance is almost as evil as AI is.

So, it's fine if you don't review it. I think the outcome is already a foregone conclusion. Even if you approved it, I don't want my IP to contribute to the surveillance and data aggregation economy.

I did not personally lead this case, but I gave the go-ahead when I was asked for a final review. From the looks of things there is no serious misconduct here. First notes started at 2025-09-18, and I was asked for a second opinion on 2025-10-01.

This is a public space and the case isn't being relitigated so you'll have to forgive me but I will need to be very careful about what I am willing to share.

A lot of our heuristics come down to things we've naturally identified by browsing specific sites which are only for AI-generated art. This means that there is very little guesswork needed, because a lot of AI artists will willingly share their methods and prompts instead of forcing us to painstakingly investigate a couple million images.

If a certain species with a certain face is in a certain pose 50% of the time, and a specific artist also draws this species with this face and this pose 50% of the time, it logically follows that they are probably using a similar basis for their work. Of course, most prompters have at least enough variety to make this 50% figure impossible, but the central point remains: it is troubling if an artist's single image resembles several thousand different AI generations, especially if the AI gens come first and not the other way around.

We do allow for counter-evidence to be submitted, but anything except a video made to our requested specifications technically allows at least one form of falsification: a worse video can use Pureref to secretly trace an image that is only visible to the artist, a .psd file doesn't contain much information about the actual creation process so an artist can trace an image and then make up a sketch after everything's done, things like that. If we forego a request for a video, it would basically only because of one of two things: the new timeline makes the artist's AI usage impossible (e.g. showing us an alternate account with perfectly similar art from 2020), or they have been keeping high-quality proof from the get-go.

In Hyena Beans's case, every submission to e621 triggered multiple different red flags, sans one. Absent any additional evidence, it would be foolish to argue "well, every other post is obviously fake, but maybe this slightly different one is real". Furthermore, we have a list of about 40 other suspects (some already handled, some pending continued investigation) who have very similar art styles despite no known link to Hyena Beans. Imagine if 40 different people all drew like Tom Fischbach despite not reading TwoKinds. That is the kind of suspect that Hyena Beans is.

post #2294835

Back in the infancy of AI art, before we had a lot of the more advanced detection techniques, I accepted an appeal that consisted of one PSD: basically everything about the artist's image and technique was perfectly explained within. I even went back and wrote a comment in their defense.

AI frauds have gotten better over time, however, so we have started to need more and more vigilance. Gone are the days where "Daggerashi" would send me their yiffy-e18 img2img inpaints to try to prove that they were real.

At the same time, needing to be suspicious of every single post would feel completely awful. If you weren't careful you could probably develop a self-induced paranoid personality disorder by just constantly scrutinizing every image and every artist comment.

I have an anecdote from our Discord server: I posted a SFW image of Haishima, a VRChat avatar, where a customized version of her with different clothes, different fur colors, etc. was sitting in a Japanese-style apartment. I didn't think too hard about it because you actually can customize Haishima to look like this, and the composition itself was massively un-AI-like, but a couple of people pointed out things like the number of fingers changing between different versions of the same Haishima. The person who made the image was doing img2img prompts of a VRChat avatar after having already customized it.

I was livid. I let down my guard for one second, thinking the image was safe because of its obvious use of VRChat materials and a very un-AI-like style, but nevertheless I was the one sharing AI art that day.

This kind of vigilance is draining to constantly maintain, and it turns a bunch of people into complete assholes over the ability to make more money with less work. I had post #2309174 favorited in July 2020 because I liked the artist's work. Fast-forward to September 2025, they give me a .psd with a layer that is 99% AI-generated coloring and I come to the obvious conclusion. For half a year since then, if anybody else posts about me on Twitter this artist often pops up to go "YEAH FUCK YOU LAFCADIO!"

There is no option that allows me to make everybody perfectly happy without giving away the game, and because from the outside looking in it appears "the man" (me, an authority figure) is keeping "innocent artists" down (artists for whom we have a significant body of proof that we can't share without either A: reducing our effectiveness by giving a blueprint to avoid detection, or B: breaking confidentiality by sharing the artist's PSDs/photos/etc.), certain people are just inclined to immediately believe the artist instead.

AI image generation and its consequences have been a disaster for the human race.

Ok, I think this has run its course, so I'll just tie off a few things.

It's worth remembering that we get the same questions about the same people multiple times - I've had to comment on AZoomer 4 separate times in forums & DMails; this case is especially good for showing how easy it can be to fake proof - & we can't customize responses every single time; it's best to check their wiki page for information on the matter first before drawing your conclusions.

It's also worth remembering that artist's complaining about us can frame the story however they want, & it's very easy for them to make it out like they did everything they could have when they didn't even try; if they're already lying about using AI, why wouldn't they lie about proving they used AI?

spe said:
I find it funny that she says she was "given a requirement to provide something that is physically impossible" - all we asked for was a screen recording. Many real artists have been able to provide this in the past. That’s FAR from an impossible or even unreasonable request[...] Deerwalker is just trying to save face at this point.

That said, there is absolutely no malice or ill will on our part. If Deerwalker still wishes to prove her innocence, she is always welcome to submit that screen recording we asked for.

I personally despise relying on "trust me, bro", but that goes both ways; we're working to increase transparency, but keep in mind that artists we've deemed to use AI have every motivation to frame themselves in the best possible light. A general healthy skepticism, in all matters, is a good thing to have.