4 ms·
This article has some examples of output from poisoned vs clean models. https://arstechnica.com/information-technology/2023/10/university-of-chicago-researcher
by jpk 3y ago
This article has some examples of output from poisoned vs clean models.
https://arstechnica.com/information-technology/2023/10/university-of-chicago-researchers-seek-to-poison-ai-art-generators-with-nightshade/ https://arstechnica.com/information-technology/2023/10/unive...
- JohnFen 3y agoInteresting, but that doesn't really answer my question. Has anyone done wider studies about the effectiveness of this sort of thing generally? As in, does it have to be tailored to a specific model or does it offer generalized protection with all models? The Ars article doesn't talk about this aspect. If these methods are effective, can a similar thing be done with text?
- jpk 3y agoI can't speak to broader study of this, but I suspect this is only doable with images because you can make meaningful changes to the pixel data that remain mostly imperceptible to the eye. I think there's less room for this sort of thing in text. Happy to be proven wrong, though!