They claim they may have a way to remove Glaze and Nightshade, two of the tools protecting images hosted on Cara from being used as training data. Which means right now they don’t have a way to remove it, which means Cara is still working, which I’m not implying is going to forever be true nor am I particularly informed; I just read the whole post.

Does CARA implement protections on scraping? Or is it meant to be a platform where even if scraped, the images are protected via Nightshade and other tech?

An individual being able to specifically target an Anti-AI site is different that crawlers being automated to do the same thing to every site ignoring robot.txt files etc.

So it took an individual to go and specific target this platform which I think means that CARA isn't really breaking its promises so much as it just needs to continue to update the techniques it uses to protect the artworks housed there. Web security has always been a game of cat and mouse.

From reading a bit more through various linked posts, I don't think they downloaded the artwork for the claimed $10, but scraped the site for a list of URLs from which to download it.

There are a number of references to the "dataset" (implied to just be this list) becoming useless if Cara were to implement stricter rate limiting.