all InfoSec news
Circumventing Concept Erasure Methods For Text-to-Image Generative Models. (arXiv:2308.01508v1 [cs.LG])
cs.CR updates on arXiv.org arxiv.org
Text-to-image generative models can produce photo-realistic images for an
extremely broad range of concepts, and their usage has proliferated widely
among the general public. On the flip side, these models have numerous
drawbacks, including their potential to generate images featuring sexually
explicit content, mirror artistic styles without permission, or even
hallucinate (or deepfake) the likenesses of celebrities. Consequently, various
methods have been proposed in order to "erase" sensitive concepts from
text-to-image models. In this work, we examine five recently proposed …
concept concepts explicit general generative generative models image images mirror permission photo public text