The media in this post is not displayed to visitors. To view it, please go to the original post.

For Instagram, my art is "Likely made with AI". Hard to be an artist in 2026...
in reply to George B

@gbargoud Shit you're right! 🤯

It's YET ANOTHER AI BUBBLE GRIFT! 🤯🤯🤯

BTW I think there's a case for fraud with these companies. Since they cannot guarantee whether something has been made with AI or not, they cannot be used for evaluation either.

It's all bullshit!

in reply to David Revoy
I put some of my works, even actual scanned pencil drawings that were never touched by a painting software, through some AI detectors just to see what would happen and my traditional pieces were often rated as being likely AI.
Esta entrada fue editada (sábado, 18 de julio de 2026, 12:55)
in reply to Takiro 🎨
@Takiro that raises the question: what are the criteria for tagging a picture as ”not likely AI“?
(Respectively, on what basis is the likelihood for ”not AI“ computed?)
🤔
@davidrevoy
in reply to Irina
@irina @pafurijaz The catch-22s many artists face. When you say, "Get off of big-tech!", you are telling them to cut their possible audience. When you tell them to sue big tech, you are putting their money for legal fees against big tech's money.
#FuckedUpWorld
OTOH, there's my prior construction boss's reason for no advertising, "If you're good, all you need is word of mouth to get business."
in reply to David Revoy

Aidan Robbins, a guy on YT who makes really beautiful mini nature/environmental documentaries made a thoughtful video on the subject on one of his secondary channels

Essentially he describes a growing intent to leave in imperfections in his work, and include "behind the scenes" type stuff to sort of make it easier for people to know it's not AI.

Not the exact same kind of art you make, but maybe some principles that might be food for thought:

m.youtube.com/watch?v=ftoUfp709UQ&pp=ygUQYWlkYW4gcm9iYmlucyBhaQ%3D%3D

in reply to contrasocial

@contrasocial If the trend catch on and enough content creator start doing it, this will be fed in the training dataset of all models and they will copy those imperfections and behind the scene stuff too.

As long as those models have free access to everything ever created, there is no escape. It's unfair competition

in reply to Doblins`

@linsorld
There's a limit to what they can emulate because LLMs cannot understand context.

When you, for example, create a "behind the scenes" style aside for something you made, it's from another perspective but is 100% consistent with the finished piece.

An LLM might learn to insert a "behind the scenes style" piece of content at a certain spot, but because it cannot understand context linking the two there will always be inconsistencies that are obvious to any human critic.

in reply to maxlath
@maxlath @me put som HG. Wells through and see if that dude had time travel after all. If he was using AI we know he could travel through time, meaning it was all made up, so he didn’t have…shit this might be one of those paradox things we should avoid.
in reply to David Revoy
I'm starting to suspect there are bots out there accusing things of being AI that aren't AI with the explict purpose of blurring the lines between AI and not AI... 🤔
It's classic propaganda tactics when no one can tell the truth from a lie the propagandist (in this case AI) has won 😒
in reply to David Revoy
as another artist i know how hard it hits you :BlobhajSadReach: just the other day some artist friend got questioned whether they used ai to draw their new art, cuz the questioner didn't understand the structure of their creative outfit, they were sad and upset just like you.. :blobugh:
in reply to David Revoy

Only way to get rid of AI is to take the money out of it. Meta is making money on this and if they haven’t already started I bet they will start making their own AI accounts.

(Which is why I think free open source AI would grind the AI companies to a halt, because there would be no money in it and then it could be made ethical for whoever insists.)

in reply to David Revoy
: the problem is that now you can find genAI that creates timelapses. I don't use them so I have no idea how good they currently are, but even if they're bad they still are enough to discredit timelapses entirely. Soon there will be no way to actually prove your creation is manmade and it's TERRIFYING.
in reply to David Revoy

i wrote a whole rant about it with all my lungs

essentially:

the problem of #AI is not fakes (although that is a problem, it's not the main problem)

the problem of AI is that it enables this lowest common denominator asshole going "fake, AI" under EVERY GODDAMN NEWS ITEM AND ARTWORK

that lazy behavior becomes a #disinfo vector all by itself

the point of the #propaganda routine the Firehose of Falsehood is to make us distrust everything, even the real

mastodon.social/@benroyce/115435710841470406

Esta entrada fue editada (sábado, 18 de julio de 2026, 14:55)

Space Catitude 🚀 reshared this.

in reply to David Revoy

This is happening to almost every creator. A lot of those accounts are AI bots, off to sow confusion and doubt by claiming any real art - real creativity - is AI.

The fact that AI is trained on our stolen works means that it will detect our creative efforts as AI... and therefore makes AI useless for detecting AI.

The tech bros who love AI want to - very literally - destroy any real creativity in favor of their fake slop.

Then they pretend to be offended when we say we hate AI.

in reply to David Revoy

For similar reasons, I am skeptical of all the people who say "You can always tell," and then pick apart other people's works looking for clues.

It's so demoralizing to be an artist and have to prove that yes, you actually did put in hours and hours of effort into something.

in reply to David Revoy
the irony/absurdity of this is that the AI used to detect AI production probably had ingested your work at some point of its training from its base model and now consider anything remotely resembling your work as "possibly" made by AI, which strangely make sense as someone could have used the AI to copy your style. A bizarro word kind of sense; the very stuff that stole your work is now used to label your original work as fake. That's just...great. 🤦🏾‍♂️
in reply to David Revoy

Whaaaat? Boooo. 😠 At least Gemini secretly watermarks everything it makes, so it knows for sure.

That point someone mentioned in the comments about your art being used to train it sounds like a viable one, honestly.

This is exactly why I'm trying to learn how to document stuff, so I can do exactly as you did. I'm pretty worried about having my reputation ruined by a false accusation of using AI. (I do like your use of Github to track the changes!)

in reply to David Revoy

"People don't believe us when we say that AI art is just as good as human art. Let them believe that human art is made by AI, until they stop believing their own senses".

Aka. gaslighting.

Esta entrada fue editada (sábado, 18 de julio de 2026, 18:26)
in reply to David Revoy

Why am I getting a vision of Pepper submitting a scroll for a assignment and the professor declaring that this is the work of Avian Intelligence, and not her own work?

I think Meta are having trouble determining what is real anymore because the well has been poisoned by so much undisclosed fake material.

Either that, or they've got a deal with Adobe to "peek" at work in progress, and they've overlooked that people often use tools other than Illustrator to do graphics.

in reply to Stuart Longland (VK4MSL)
@stuartl The idea the Adobe is spying on users to power "AI detection" is the sort of underhanded, deceptive shit that is so on-brand for Adobe. 🤣 My eldest is a visual arts major, and complained a lot last semester when her 2D design class requiree her to use Adobe Illustrator.
in reply to VANTABlack

@VANTABlack @David Revoy Maybe not even that.

Given the amount of AI-generated “slop” out there nowadays, statistically, the likelihood that something is AI-generated is becoming greater and greater.

So maybe they don’t try to actually detect AI slop, and instead just do something like this:

int is_ai_generated(void *stuff)
{
    // Don't bother looking at *stuff; statistically,
    // 90% of the stuff is AI-generated anyway
    if ( get_random_number(0, 1) ≤ 0.9 ) {
        return TRUE;
    } else {
        return FALSE;
    }
}

(I’m only half-joking: I wouldn’t be surprised if that’s basically what many “AI detection algorithms” actually do.)
Esta entrada fue editada (sábado, 18 de julio de 2026, 22:35)
in reply to David Revoy
I am so sorry that such wonderful work is a victim of rubbish AI nonsense. How can this be fixed? False automated accusations are actually libelous and I would take it on that way. There must be a artists' association that can so this? Good luck. Great paintings.
in reply to David Revoy

Well, now I follow you there, it didn’t even came to my mind before ^^ (there, I am rather getting news from a few non Fedi fellows, and about people around nature, ecology, ethology, plus rather independant news media....)

I am sure, our best strength is to enforce a collective human effort to make human creation on the foreground of the scene. I don’t know if it could be better, if there are drawbacks, but I am more and more attracted by codeberg.org/robida/human.json.

in reply to David Revoy
I got kicked out of a high paying freelance gig because my organic writing kept tripping their "industry standard" "AI detector." Said (alleged) detector was completely paywalled and just gave a percentage match, didn't highlight phrases, so I wouldn't have been able to fix individual issues. I suspect part of it is that I write like an autistic person, and another part is that when all you want is soulless advertising copy, it's impossible to distinguish the source.
in reply to David Revoy

Sensitive content

Esta entrada fue editada (jueves, 6 de agosto de 2026, 8:50)