Stop-Ai-replace

If Core ever expires:

Welcome to the Stop-Ai-replace page, where I serve to resist, teach, about the abundance of AI images!

Feel free to click on the links

• How to block any AI-generated 'art' - a tutorial on how to scrub out the large amount of AI images from your timeline

• AI 'art' account free blocklist - tracking any AI account to avoid and block

Tools and software to resist or detect AI:

• Glaze Project - cloak your images to make AI harder to mimic.

• Illuminarty - submit the image to detect the percentage of image's AI probability

I do also post about news in the Post section.

I make announcements, news, deviations, and more, harvested from the outside of the DA (with credit). You can watch if you want journals about AI from me. I guarantee to watch back.

I pop up frequently in the comment section of deviations to be a simple indicator of the piece being an AI-generated image. Some users aren't willing to read tags (or know the machines featured in this case), descriptions of tools they're using, or etc. A lot of AI images won't tag or say in descriptions that their stuff AI nor the specific generators they're using.

Feel free to expand my bio for solutions and entire paragraph how the AI machine work and argument against it. You can also check DA's forums in '...' button how users talk about plague of AI-generated images spread and praised for it.

Let's stop this contagious AI phase once for all and protect artistic property.

Before you submit the ticket, please take expand and look into paragraph down below and gather information.

Our voice can be heard if we submit ticket/contact to support. Spreading this word with other people  (via deviations, chats, forums, etc) will make the deal.

Product Feedback and Feature Requests - DeviantArt Help Center (deviantartsupport.com)

(I ENCOURAGE YOU TO WRITE THE CONCERN CIVIL TO TAKE IT MORE PROFESSIONALLY.)

Thanks to Inafoxy for wonderful information!

------------------------------------------------------------

Here is public model proof that unlicensed content is being used in the training model:

https://laion-aesthetic.datasette.io/laion-aesthetic-6pls/images?_search=artstation&_sort=rowid

Diffusion-style AI models don't "learn" they use signal-noise aberrations to mix photo content, the composition IS the images. The semantical processing just creates contextually-learned placement on a diffuse grid which is only as many rearrangements as the text prompt gives. Everything else is composition from the diffusely stored images, wherein it is "similar" distance lookup (hence how loose-drawing prompts are possible). These AI imagen gen models are just glorified similar-reverse image searches, which is nothing like a real artist learning to understand or learning art, especially since the semantical composition processing is higher order and algorithmic. AIs are limited to their algorithm and the pre-trained images, remove that and it doesn't work. Copyright images don't belong in the AI without permission, at both the fore-training level and the compressive analogous-encryption level. Unfortunately people who don't understand how an AI works might not even realise this when using it.

Just as with anything compressed or channelised just as in WiFi transmissions, AIs actively storing data in an effectively classically superpositioned state is theft. If this is acceptable why not scramble the nuclear launch codes over a diffusion based AI while we're at it? The AI is literally storing the essence of the hard work of the artist, this is merit/credit abuse and "similarises" images in a similar way to the process more primitive methods like MPEG-4 IPB interframe compression uses. The AI can literally reproduce the original content as it was in respect to its intellectual properly, if an artist tells a specific message through their art and an AI can reconstruct that then we must start perceiving these AIs as a form of encryption rather than as if the AI itself is an artist. A real human artist can't reconstruct the semantics of an image without tracing from its crucial components, yet AI frequently does this wholly or partially. This is an encryption nightmare as much as a copyright one. You wouldn't visually based private messages in an AI diffusion model, it's not a true scrambler and is just slightly lossy. It's enabling IP, aesthetical and textural labour theft for the purposes of monopolism and consumerism.

Share these concerns on your artworks and anywhere else you communicate with others.

--END OF POINT--