PLA: Prompt Learning Attack against Text-to-Image Generative Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lyu, Xinqi, Liu, Yihao, Li, Yanjie, Xiao, Bin
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912521885057024
author Lyu, Xinqi
Liu, Yihao
Li, Yanjie
Xiao, Bin
author_facet Lyu, Xinqi
Liu, Yihao
Li, Yanjie
Xiao, Bin
contents Text-to-Image (T2I) models have gained widespread adoption across various applications. Despite the success, the potential misuse of T2I models poses significant risks of generating Not-Safe-For-Work (NSFW) content. To investigate the vulnerability of T2I models, this paper delves into adversarial attacks to bypass the safety mechanisms under black-box settings. Most previous methods rely on word substitution to search adversarial prompts. Due to limited search space, this leads to suboptimal performance compared to gradient-based training. However, black-box settings present unique challenges to training gradient-driven attack methods, since there is no access to the internal architecture and parameters of T2I models. To facilitate the learning of adversarial prompts in black-box settings, we propose a novel prompt learning attack framework (PLA), where insightful gradient-based training tailored to black-box T2I models is designed by utilizing multimodal similarities. Experiments show that our new method can effectively attack the safety mechanisms of black-box T2I models including prompt filters and post-hoc safety checkers with a high success rate compared to state-of-the-art methods. Warning: This paper may contain offensive model-generated content.
format Preprint
id arxiv_https___arxiv_org_abs_2508_03696
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PLA: Prompt Learning Attack against Text-to-Image Generative Models
Lyu, Xinqi
Liu, Yihao
Li, Yanjie
Xiao, Bin
Cryptography and Security
Artificial Intelligence
Computer Vision and Pattern Recognition
Text-to-Image (T2I) models have gained widespread adoption across various applications. Despite the success, the potential misuse of T2I models poses significant risks of generating Not-Safe-For-Work (NSFW) content. To investigate the vulnerability of T2I models, this paper delves into adversarial attacks to bypass the safety mechanisms under black-box settings. Most previous methods rely on word substitution to search adversarial prompts. Due to limited search space, this leads to suboptimal performance compared to gradient-based training. However, black-box settings present unique challenges to training gradient-driven attack methods, since there is no access to the internal architecture and parameters of T2I models. To facilitate the learning of adversarial prompts in black-box settings, we propose a novel prompt learning attack framework (PLA), where insightful gradient-based training tailored to black-box T2I models is designed by utilizing multimodal similarities. Experiments show that our new method can effectively attack the safety mechanisms of black-box T2I models including prompt filters and post-hoc safety checkers with a high success rate compared to state-of-the-art methods. Warning: This paper may contain offensive model-generated content.
title PLA: Prompt Learning Attack against Text-to-Image Generative Models
topic Cryptography and Security
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.03696