I decided to do a bit of digging and found an account on Xitter called @toyxyz3 that's pro-AI. He and one of his followers went over the methods being used to "protect" art from AI training. Apparently, none of them work. Techniques like Nightshade, Glaze, Grains, Gaussian Blur, Watermarks, and Noise can all be filtered through existing methods. While these techniques may make the AI-generated image look a bit off, it can still end up looking better than the original mess that the artists posted. Artists are purposefully ruining their own content for no reason.
Discussion
To be fair, he does state that it's possible to "fool" the AI by making increasingly intrusive modifications to your pictures. In my opinion, if you're going to share a picture like this, you might as well not share it at all.
The cope that "it's just img2img" is stupid too. Entire models have been trained using Glazed images, Nightshade most likely can go through a similar process. And again, we already have enough data to imitate just about any artstyle.