I know this is a long shot but for those who don’t know about these 2 amazing tools Here’s the official site Glaze protects work Nightshade poisons AI models data is how I’ve had it explained
I’ve seen concerns about bots scraping data from Fediverse users and I’m concerned as well. I want to know is there a way we can protect ourselves or a way we can fight back?
In theory I don’t think this is feasible for any text that can be read by robots (eg the fediverse)
If the file containing the text is itself poisoned then maybe?
Unfortunately, Nightshade can already be thwarted by AI.
I have seen those who have left reddit put a word scrambler to work on their accounts, editing all of their comments into garbled nonsense. I’m not sure how effective that is to deter AI scraping, especially if reddit can just pull from history before the comments were edited. Also, it is a more destructive means of obfuscation because your comments will be unreadable to humans as well.
I do hope a solution comes along, though.
Unfortunately, Nightshade can already be thwarted by AI.
Do you have concrete proof for that? Because it’s literally what they want you to think so you don’t bother poisoning your work so they can have more free data
It appears that LightShed reverses Nightshade and Glaze with 99.8% accuracy:
https://cybernews.com/ai-news/glaze-nightshade-cant-stop-ai-scraping-art/
Point taken, thank you
Attach the entire pdf of codex seraphinianus to all your texts like some ancient-nomadic child putting wolf teeth under their pillow to protect from the evil eye.
