Forums » Suggestions & Development Discussion » Gen-"AI" warnings?

Would it be possible to add a warning similar to the violence/sexual/language content to characters or even user profiles for that would label computer generated images or writing? Like plenty of other creatives, I strongly prefer avoiding "AI" imagery and find generated text as a reply as unfun (it's also worth mentioning the breach in trust of a partner entering writing that isn't theirs into the generator)

I think this responsibility is on the user. I identify all A.I generated art I use, “Face Credits", or commissioned art for my characters.

As for writing, I think people need to be upfront and talk to their partners. Even though I consider myself fairly fluent in English, it is a second language and I use ProWritingAid to help correct my writing in both French and English.

I suggest communicating with the people you are interested in writing with.

If the way they speak in OOC text, spelling, vocabulary level, syntax, etc, differs from how they write IC, you could address the issue politely.

As for art, I usually respect other people’s preferences and will change what I use if it really bothers the other person, and I feel I can accommodate the other person without sacrificing my vision of what I think represents my character.

It comes down to honesty. If the option were there to click similar to other warnings, some would and some wouldn’t, just as some people identify where the art they use comes from and others don’t.

This was a suggestion or question about possible feature implementation, which is why I didn't feel the need to go into detail about my reasons for not wanting to see it. I'd rather stay on topic and not steer this into an argument about the benefits of LLM use or why specific cases might make the suggested feature not needed for you specifically.

I personally wouldn’t mind an AI-warning. I admit using it for some pictures, like Lizbeth’s icon picture or to make up some nice pics for the role-play sessions.
But it will be optional to show that there is some AI generated content. As using free AI isn’t regulated by any law, no copyright, no commission, no rating. So it will only be a personal warning or signaling.
An idea would be a AI-free certification, like verified adult, a verified human (not only through ‘are you human’ tests). Like I would be 80% AI free, human written text but occasional use of AI for illustration and proudly show that on my profile😊

I tend to side with AutumnFox in that it is the responsibility of the user to identify their interactions with AI.

Since there is a mention of the 'violence/sexual/language content' warning settings; there is already a 'custom' warning selection where one can write text identifying further warnings, such as AI use. Of course, all these warning ticks are subject to the character creator using them so I don't think it would guarantee you not running into AI imagery and text.

Yeah, I'm not sure how many people would use these warnings if they became a feature. Even if users were required by the site rules to warn about AI usage, they could lie and deny it, and it probably couldn't be proven one way or another. I've certainly seen writing here that looks AI-generated, but that doesn't guarantee that it is and if the user says it isn't, we can't really prove whether they're being honest or not.
For the same reason, a rule outright banning Ai-generated content probably can't be easily implemented/enforced, either (though correct me if I'm wrong about this), as anybody could just lie and say they created it themself.

I could see such a feature acting as basically a reminder to be upfront about the use of generative AI... for those who would bother to use the label, assuming they even had the same (or similar enough) understanding of what it meant as you do with your request or Kim might in writing it out. Like... there have been tools I've used in the past that I now think might count as generative AI, but I'm not sure? Similar technologies were used in some cases, at least, but with UI or setup that made actually using it seem much more like some other tools I've used that I'm pretty certain would not qualify as generative AI. So it can be kinda confusing sometimes, to be honest.

Not to mention all the people who just grab images from around the internet without any idea where they actually come from.

So... maybe a feature like this would have enough of an impact to be worth the effort to implement it, and maybe it wouldn't. Things needing the existing violence/sexual/language markers can at least easily be identified and corrected when they're missed, but the only things I can think of to help catch things missing an AI-gen label would require possibly quite a bit of extra programming and still seem like it'd give a lot of false positives and false negatives.

And I think there's a lot of people who simply would intentionally not click the box, for various reasons that I don't think are worth listing out here.

If I thought that a 100% effective solution were achievable, I'd probably be bringing up discussion on whether generated text or imagery has a place on an RP site at all, but that's not what I'm suggesting.

I don't find the argument of 'some people don't use the labels' or 'some people lie' convincing. Both can be said about current features or rules of RPR. Conceptually, the idea is no different than filtering out character profiles marked as containing offensive language or sexual content, which is already a feature. By the same argument repeated in the replies, there's not much point to the sexual content warnings because people might decide to ignore them. The custom warning isn't really a viable solution, as it does not suggest labeling any specific content and no site-wide precedent for it has been set.

I'm fine accepting a moderator or Kim shooting down the idea and won't argue. But I'm a little weirded out about how many people here commented to say how no one would care about such a feature.

The content warning checkboxes (outside the custom one, of course) are moderator enforced and can be reported if not used correctly, so I don’t see it being too much of a stretch to perhaps have one for either/and profile/gallery images specifying there was AI used. That way people could perhaps filter those profiles out or toggle visibility as a whole.

I like this idea a lot. I don't think I can implement it immediately but it's definitely going into the hopper.

At the moment, I've just decided to start blocking users who use AI slop. Would be nice to get a warning so I don't have to find them outright admitting it or trying to puzzle through their images and text. I'm here to be creative, not play with a computer. Self labeling would save me SO much time - as I'm sure they don't want to deal with me and my 'elitism'.

Pinterest now has an ‘AI modified’ disclaimer at the bottom of images, which are about 88% of them these days. Adding an ‘AI’ tick box warning on the character profile will be a nice feature, but also make it a search function where results can be excluded for those who wish not to see AI characters, or included for those who want to see AI characters. My experience with roleplaying with someone who uses AI generated computer art led to them eventually having others generate their replies. That isn't much different than having AI generate replies. At the end of the day, it's my responsibility to identify such text content. The graphics content is usually pretty obvious.


What colour of post-it note will be stuck on your computer monitor I wonder?

I love this idea. I personally have zero interest in RPing with anyone who uses generative AI in any capacity, and while I've got a really good eye for the tells, this would be very helpful. I'm really glad to see that Kim is on board, and I wanted to weigh in with my support for it, too.

Thank you for pitching it!

I would appreciate this too a lot. I would like to at least have a fair warning!

Ok, maybe I'm misunderstanding here but I have a lot of concerns...

Now, I'm not a huge fan of AI, but this site is really more about writing than art. I can understand not being allowed to post artwork created by AI in the art sections, but I have a number of questions about if this is going to be a thing on people's profiles. I completely understand trigger warnings for violent or graphic content. It is essential for sexual content because there are actual laws about minors seeing adult content (and I 100% stand behind this).

That being said, here are some things I'm wondering about if this is going to be a mandatory warning required on someone's character profile:
- How or why is it anyone's concern, especially if they are not roleplaying with the character in question? How are you harmed by it?
- Are we talking feature portrait or if you have any AI images in the gallery widget on a profile?
-What constitutes AI exactly? Well before Chat GPT, there were avatar generators, a quick and free way to put a simple picture together. Is that AI?
-For years, a common way to represent a character, particularly in modern day/slice of life is to use a real-life celebrity picture to represent your character. How is this different from using an AI at the end of the day?
-Are we going to only allow artist created renderings of characters to avoid the warning label? That almost feels like extortion.

One concern I have is just for fun, I made a few additional pics of one of my character using chat GPT, especially a few group shots that I would otherwise not be able to do. Does that mean I have to have "a warning" on that profile now? This character is only used on a private one on one game, so someone else RPing with this particular character isn't really an option here. Another example, I made an "action figure" of one of my characters and posted it to her profile. Why does that need to be labeled as such? I used an app specifically designed to do action figure mockups. Why does that need a warning? How is someone being harmed in that instance looking at an image of an action figure created by an app? Does that app fall under AI or is that not the same thing?

Don't get me wrong, maybe I'm completely misunderstanding here but hoping someone can clear this up.

I prefer not to RP with characters that have AI images or writing involved in their creation or writing, so I would appreciate the ability to filter out characters that have a "created with the use of AI" tag.

I usually can tell when AI has been involved, but sometimes not. Plus, it would simply save time by not seeing those characters to begin with.

Looking at the user above's argument, this seems to be a "right to know" issue. I would make the argument about underage users (minors). I am adamantly against doing an RP with a minor with the exception of things like Halloween and other one-off site activities.

But posting age isn't mandatory. I feel that is far more important, however, than whether or not an image on a character is AI or not. That being said, I also understand that that becomes a privacy issue, too. If you mark everyone under 18's account, that targets them and becomes a privacy and safety issue with predators.

So at that point, it makes it my responsibility to ask the perspective partner who doesn't list their age on their profile, are you under 18? Not sure why if this issue is that important and you have questions about the sources of someone's character icon you can't do the same. I would assume you'd look at their profile and then if you aren't sure you can ask them. If I'm going to reach out to someone about doing an RP, I generally start a conversation to see if we're on the same page.

That also brings up the issue of open RPs. Take the one called "Trixie's Bar" that has 90123851236416 replies because it's been going on for years and is still active. At that point, if you are joining a public RP, then someone is going to have some AI art anyway, regardless of them having a disclaimer on their profile. It's going to be impossible to avoid AI art on anything not private. So again, if you're looking to start an RP with someone privately, that seems like something you can ask them.

Generative AI raises an enormous amount of social, ethical and moral questions. There are a lot of people who have examined the problems with it and decided that they want no contact with it, or at least as little as the world will allow them to have. I am definitely among them, and the number of problems coming down the pipeline for the human internet at large because of AI are staggering.

I could also ask, how would you be harmed by a note below an image that says "AI generated", or some kind of way for people to be warned before looking at profiles that contain AI generated text/images for people who want that? The site already makes it easy enough to toggle whatever warnings you like on/off.

If you aren't looking to pass off AI generated art or text as something you personally created, I don't know what the problem would be with being clear about the source.

I imagine that much like now where you can mark an image in a gallery mature without having to mark the entire profile, we could do it image by image unless you've built the image directly into the profile, or if you are using ChatGPT or similar to come up with your character descriptions/history/etc.
Not even remotely. Avatar builders were large collections of assets designed by a real artist, which could then be mixed and matched, and present none of the same social/moral/ethical quanderies.
Photography presents none of the same society-wide social/moral/ethical quanderies. It was also, up until recently, pretty easy for people to know that they were looking at a real picture of a celebrity and to make a decision based on that knowledge about whether they wanted to engage with that character.
All options other than AI are still on the table, exactly as things were just a few years ago.
Or at least on that gallery image.
I think we need to know what the app actually was. Did you use an app designed for mocking up action figures, generative AI, or a skin on generative AI that only does action figures? If you're talking about something like hero forge, that's basically just a 3D version of the above-mentioned avatar creators that hosts assets made by real artists that can be mixed and matched.

The big question to address here and why I feel that way is the label feels like shaming or even bullying. I see it as a scarlet letter. While everyone has been respectful of stating their anti-AI opinions, it feels like those people who are forced to add this warning are going to be shunned or outcast from the group, thus the feeling of bullying. Polite or not, several people have already commented about essentially blacklisting anyone that has anything AI on their profile. Not exactly welcoming or inclusive is it?

Again, I use very little AI on my profile and I do see some validity in some of the concerns, I just don't see it as the same kind of issue as adult content or potentially upsetting like foul language or violence. Well intentioned policies often have unintended consequences. Does a policy that isolates even a few people that have been long time supporters of this site or drives them to leave because they suddenly feel judged seem worth it?

This site seems to pride itself on being a safe space and inclusive. I think it's fine to voluntarily offer that warning, but if it becomes mandatory, it feels like that's starting to impose rules resembling an HOA. Home owners associations started off with good intentions, too, but an awful lot of people feel they are misguided and lead to what feels like bullying or over-reaching in some form. I'm all for rules that feel reasonable and solve a problem- this one... well, I think I pretty well laid out my argument above.

Jumping to your last question about the app I used- I don't know. I find all sorts of little tools I enjoy messing around with. I'm almost 50 years old and not that tech savvy, honestly. I can't remember what tools I was messing around with months ago. That brings up a new issue- does that mean I have to go back and find such tools and determine if they meet standards as to what is or is not "AI"? That seems like an awful lot of work.

I consider myself a writer. I don't know a lot about graphics and apps and art tools. All I want is visual depictions of what my characters look like. It feels like aside from finding a public image of a celebrity, I'll either have to leave my icon blank to be safe or feel guilted into hiring an artist to draw all of my characters if we have to start applying AI labels.

Don't we have enough going on in the world dividing us as is without bringing this into the mix, too?

I feel that HOA comparison is extremely disingenuous.

Some people can upload NSFW or triggering material and they would have to tag that material, and potential partners could use that to make a decision over whether they want to engage.

Similarly, if some people use AI, they should see no issue in tagging them and allowing their partners to decide whether or not they want to engage. The only reason I see that would come into conflict with this is if you were using the AI to deceive potential partners into thinking you created something where you did not.

This argument is setting aside the enormous ethical issues AI presents, but frankly no conversation about AI will get very far on a site consisting entirely of creatives without bringing the ethical issues into play.

Many players on this site are artists, all are creatives - and rather understandably they have every right to have strong feelings on the matter, and every right to refuse to play with someone who is in stark opposition to those feelings. It’s a comfort thing and nobody is owed RP with someone who partakes in behavior that makes them uncomfortable.

RPR is hypothetically providing a tool not unlike its NSFW tags; the only isolation that would be happening would be because one is partaking in the isolating behavior.