some thoughts on getting my art opted out of AI
So I complained a lot on Twitter yesterday about the process of “opting out” my artwork via spawning’s “haveibeentrained.” It’s tedious, the website doesn’t work all that well, and I’m having difficulty finding matches for artwork I’m pretty sure is in the LAION set despite doing a couple of uploads to try and match. But still, I thought, I’m annoyingly stubborn and have a bit of time on my hands, it’s worth checking this out.
For anyone not in on it, spawning is apparently an organization developed by a few artists that is acting as some kind of arbiter between the art community at large online and AI companies to clarify permissions for AI databases, either opting in or opting out. I agree, this sounds sketchy. Still, if there’s an opportunity to opt-out my work, I’m going to try and do it. Right now it feels like kind of a futile attempt. Why this is the method we’re going with when copyright law already exists, I do not know! Probably because most artists are too broke to afford a lawyer. Still: if you know a copyright lawyer out there looking for what’s probably an easy case, they should probably hear about all this.
Some of the things I found out while I was digging for my own artwork are in this thread on Twitter:
https://twitter.com/radio_chio/status/1605300054977396736?s=20&t=gxPpdsjFdrpbDuBND4wDMQ
Here are my initial thoughts on the spawning site when I started using it. https://twitter.com/radio_chio/status/1605293929594867750?s=20&t=gxPpdsjFdrpbDuBND4wDMQ
Interestingly, I got a reply from someone who works on the site, probably because I complained about the bugs. It would appear that if you host your own images on a domain, you can request that that domain be opted out in one go. I don’t know whether they followed through on this, as I’m still finding images in the set, or if they just haven’t done it yet, but it’s a tidbit of information I’m going to throw out there because why keep this to myself:
[image: a tweet from @jordancmeyer on twitter replying to @radio_chio, Also if you have a website hosting a large number of images, please send us the URL of the site ([email protected]) and your HIBT username, and we can try to flag all of the images that start with your URL in one go.]
So, I make no promises, but maybe it’s worth sending an email if you own your own site.
I’ve been fiddling around with HIBT a little bit more today, and find that I’m having ever so slightly more success with finding my own images when I phrase my searches like it’s an AI prompt. Example: “webcomic page watercolor messenger bugbyte” Cool cool cool.
Annoyingly, most of what’s coming up is stuff that’s cached in places I don’t necessarily have control over. Deviantart, for one. I tried DA’s portfolio tool one time and didn’t actually publish anything, and all my DA art is in Wix’s cache now, and that was scraped by LAION, despite not using Wix or DA portfolio since they use Wix as a platform. I used jetpack to cache my self-hosted Wordpress site, and some of my images wound up in a cache on a Wordpress-owned domain, presumably as a result of jetpack. It is a huge mire of trying to find all the ways that your images might slip into other places without your knowledge.
I don’t think it would be respected, but I really wish there were a way to just add something to the robots.txt on my website(s) and opt everything out forever. I feel like this is just the first round of what is going to be a very long and aggravating process for a lot of people.
