Asymmetric Bias in Text-to-Image Generation with Adversarial Attacks

arXiv:2312.14440v2 Announce Type: replace
Abstract: The widespread use of Text-to-Image (T2I) models in content generation requires careful examination of their safety, including their robustness to adversarial attacks. Despite extensive research on adversarial attacks, the reasons for their effectiveness remain underexplored. This paper presents an empirical study on adversarial attacks against T2I models, focusing on analyzing factors associated with attack success rates (ASR). We introduce a new attack objective - entity swapping using adversarial suffixes and two gradient-based attack algorithms. Human …

abstract adversarial adversarial attacks arxiv attacks bias content generation cs.cr cs.lg image image generation paper research robustness safety study text text-to-image type