I appreciate the thread, and the thought! I'll say that before I jump in, because I almost completely disagree with you ;).

I'm not swayed by the "if I see something I just can't get it out of my head" or that you might be tricked by an image that is generated by AI. In those cases, I don't think we've banned people in the past for having seen an advertisement or seen an image and not being able to 'get it out of their head' and drawing something similar, using it as inspiration. I can point to the pure number of items created in the CW system, before LLMs were wildly available with image generation, and that number would seem to tell a different story. If you could do it two years ago, I'm sure you can do it now by asking your commissioner for another image that doesn't come from AI (and I know you can mostly tell).

What I'm trying to protect against here is two fold:

1. Protecting the artists who've been drawing things online for 20+ years, that have had their art gobbled up by these machines. I'll be completely upfront, I'm very annoyed that you can ask most AI generators to make something 'in the subeta style' and it will generate something based on years of hard work that my team of incredibly talented artists have done. That's frustrating, and likely to make it very hard for me to accept any usage of AI in any capacity, sorry! Talk to Sam Altman, lol.
2. Our site. I appreciate that everyone wants to a team of CW folks handling CWs, but we're running with the incredibly small staff that we have because that's the amount of money that comes into Subeta. And that's on top of getting me for free! What we absolutely cannot afford is any kind of lawsuit from a copyright owner. Just being able to use anything you see on the internet, regardless of who owns it or who created it, doesn't fit in with our ethos here.

So, I'll talk to Jessi about the repercussions of breaking these rules (which should not be getting frozen) but am otherwise not swayed into changing any rules on using AI for reference images. And, like I said, we just don't have the money to hire a team of people to manage the CW queue, that's the reality of running a shrinking budget, unfortunately! And if I'm reading this post correctly that was pointed out in the first thread... it sounds like Jessi has already been willing to give the benefit of the doubt, and work within the rules.

I don't think you should be frozen for using an AI reference image, I don't think you're stealing or trying to rip anyone off. It's just not a thing we want to be used as reference images for things we're paying to have uploaded to the site. Going back to , we have a legal liability on top of the fun and magic we're trying to create here.

And, we are unlikely to have any additional folks to handle the CW queue. It's a thankless, long job that we wouldn't put on any volunteer, which means hiring someone additional. We don't have the funds to make that happen, and are unlikely to. It's a bummer, but that is the reality, right now.

If the rule is too onerous and we see a freeze out of artists creating CWs, or have an increase in sending back items because of AI references I'd be more willing to entertain that this is a problem, but that just doesn't bear out in the data.

I don't understand what you're asking for. We've said we aren't interesting in banning or freezing if you do end up doing that by accident, and are willing to give a significant amount of leeway. It is not that much to ask for the artist go back and ask for additional, non-AI, references, or am I misunderstanding?

Can you be specific as to what rule you're looking to change or create.

And, for every one on this thread, arguments are incredibly tarnished by "if you don't do what I want, i'll stop spending / spend again for the first time in years". It doesn't come off as anything but hostile blackmail.

I really do appreciate the feedback 🙏. I'm asking for answers (and more questions) so that I can get an understanding.

If what we're looking for is an increase in specific examples and more rigorous rules so that no one feels like they're going to accidentally fall on the wrong side of the rules - I completely agree! I'm happy to work with y'all to iron out pages of explicit examples, but what I'm getting from this topic is that some folks would like the rules to be changed in terms of what is an acceptable reference (AI generated art). Your examples of things like real lines of clothing don't fall in that same area, to me, because that's fair use.

We're talking about two different things, a known and explored area of copyright with things like fair use having been established over many many years, and this sudden and incredible technology that no one has a firm handle on. Like I said, I'm particularly frustrated that tens of years of work, including the custom wearables we're talking about here!, was sucked up into LLMs without our permission.

A few thoughts, aha. I'll be entirely honest, we've benefited from having Jessi lead this effort for a lot of reasons, but one of the primary ones is that they are from the Before Times (before CWs on the site). So, while you've all formed opinions based on someone who has always been in charge, there has been a natural separate that lets Jessi do the job.

If I offered up 5 volunteer slots tomorrow (and I'm trying to be incredibly kind and charitable here, because I love a little online drama :P) and filled them, we'd immediately start getting reports from other CW creators that someone who dislikes them is on the queue and that's why their denials have shot up (not that we'd just be processing the queue faster). If I only took people who weren't actively in private dueling CW discords or hadn't spoken up loudly about rule changes in the past, I'd get accused of not taking it seriously. Opening this kind of thing (and not just needing to have some kind of time commitment, it is a really hard gig!) isn't the kind of thing I could just press a button for.

I think that is the misunderstanding. Jessi did say that it's a hard and fast no, and a ban, but after getting feedback like this that is absolutely something we're willing to massage together. That can look like relaxing the rule to when it's obvious or just un-noted (or adding a check box so that it can be acknowledged on the page itself, or something!) to a warning and asking for another image.

Jessi was reacting like how I was above -- with the feeling of extreme annoyance at seeing the art of coworkers be slurped up and used in LLMs without any compensation or attribution. I'm willing to say we've got a little wiggle room.

.net">keith.net (and .net">amber.net) go to us!

I promise I'm not ignoring your specific questions, I don't want to set any policy with my responses until I have a little bit more understanding which means chatting with and others internally as well as looking at the responses here.

I just want to make sure when I say something specific that it's not changed a few minutes later!

This is painfully untrue. We can block the big players (OpenAI, Amazon, Google and now Apple) through various methods that in turn block legitimate use. We can use methods like robots.txt to block AI crawlers that adhere to those rules, and then we can spend an additional couple hundred dollars a month on traffic analyzers to ensure that bad actors, who don't adhere to those rules, aren't getting through. All together the effort to 'stop it from being scraped' is weeks of my time, a monthly fee and hoping and praying apple/google/openAI don't change their rules tomorrow.

Entire companies, like the one I left working at a few months ago are scrambling to provide very expensive avenues of preventing text and images to be taken by AI. It's absolutely possible to block AI crawlers, it's difficult and expensive to do so.

I'm not sure where your point is getting to here. If that's the case, we should just throw up our hands and say "well, we'll never be able to stop this" and get rid of any AI related rules. May as well let people submit fully AI drawn images if we're not going to be able to determine the difference in short order. I think we both agree here:

Which is why I'm interested in loosening the rules and making it a warning or re-submit, instead of a full (CW) bannable offense. I'm not willing to just lose the war, though ;)

Sorry, when I say war I mean the war against our robot overlords / the AI systems that are coming down the pipeline. The war that Subeta was pulled in to when OpenAI and others created systems of massive VC funding to profit off of our work, which is a bummer! Not a war no this topic or with any of you -- I would never use language like that and I'm really sorry it was interpreted that way!

Stepping outside of this thread for a second, I left my job in May (engineering manager for a 200M~ product line) and have been focusing on Subeta and consulting since then. I'm currently consulting with a few medium/large tech companies, and working on a proposal for the federal government in using AI to process applications for things like SNAP benefits (which I'm not a fan of, but know that it's coming so I may as well have an opinion on it :)). My view on LLMs and how they're being used is shaped by how I've seen my own copyrighted material be used (in a way that i'd call illegal!!) but also how the industry is moving toward replacing low income or "low skill" positions with LLMs that aren't fully baked or capable of doing the job. All of that is to say that my vision on this topic is here in the discussion we're having about Subeta and the CWs, but also a little bit broader as well. I'll completely own that I've focused this thread narrowly on the AI question because it's something I'm very interested in!

Now for updates, I've had a chance to circle up with Gemma and Jessi, and we're discussing broad rule changes, and also where our personal lines on these issues are. Then we'll work backward from there to a rule that we can all feel comfortable enforcing. Thank you for giving us something to consider there!