Enregistré dans:
Détails bibliographiques
Auteurs principaux: Liu, Yufan, Zhang, Wanqian, Chen, Huashan, Wang, Lin, Jia, Xiaojun, Lin, Zheng, Wang, Weiping
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2510.24034
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918175189237760
author Liu, Yufan
Zhang, Wanqian
Chen, Huashan
Wang, Lin
Jia, Xiaojun
Lin, Zheng
Wang, Weiping
author_facet Liu, Yufan
Zhang, Wanqian
Chen, Huashan
Wang, Lin
Jia, Xiaojun
Lin, Zheng
Wang, Weiping
contents Despite rapid advancements in text-to-image (T2I) models, their safety mechanisms are vulnerable to adversarial prompts, which maliciously generate unsafe images. Current red-teaming methods for proactively assessing such vulnerabilities usually require white-box access to T2I models, and rely on inefficient per-prompt optimization, as well as inevitably generate semantically meaningless prompts easily blocked by filters. In this paper, we propose APT (AutoPrompT), a black-box framework that leverages large language models (LLMs) to automatically generate human-readable adversarial suffixes for benign prompts. We first introduce an alternating optimization-finetuning pipeline between adversarial suffix optimization and fine-tuning the LLM utilizing the optimized suffix. Furthermore, we integrates a dual-evasion strategy in optimization phase, enabling the bypass of both perplexity-based filter and blacklist word filter: (1) we constrain the LLM generating human-readable prompts through an auxiliary LLM perplexity scoring, which starkly contrasts with prior token-level gibberish, and (2) we also introduce banned-token penalties to suppress the explicit generation of banned-tokens in blacklist. Extensive experiments demonstrate the excellent red-teaming performance of our human-readable, filter-resistant adversarial prompts, as well as superior zero-shot transferability which enables instant adaptation to unseen prompts and exposes critical vulnerabilities even in commercial APIs (e.g., Leonardo.Ai.).
format Preprint
id arxiv_https___arxiv_org_abs_2510_24034
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AutoPrompt: Automated Red-Teaming of Text-to-Image Models via LLM-Driven Adversarial Prompts
Liu, Yufan
Zhang, Wanqian
Chen, Huashan
Wang, Lin
Jia, Xiaojun
Lin, Zheng
Wang, Weiping
Computer Vision and Pattern Recognition
Despite rapid advancements in text-to-image (T2I) models, their safety mechanisms are vulnerable to adversarial prompts, which maliciously generate unsafe images. Current red-teaming methods for proactively assessing such vulnerabilities usually require white-box access to T2I models, and rely on inefficient per-prompt optimization, as well as inevitably generate semantically meaningless prompts easily blocked by filters. In this paper, we propose APT (AutoPrompT), a black-box framework that leverages large language models (LLMs) to automatically generate human-readable adversarial suffixes for benign prompts. We first introduce an alternating optimization-finetuning pipeline between adversarial suffix optimization and fine-tuning the LLM utilizing the optimized suffix. Furthermore, we integrates a dual-evasion strategy in optimization phase, enabling the bypass of both perplexity-based filter and blacklist word filter: (1) we constrain the LLM generating human-readable prompts through an auxiliary LLM perplexity scoring, which starkly contrasts with prior token-level gibberish, and (2) we also introduce banned-token penalties to suppress the explicit generation of banned-tokens in blacklist. Extensive experiments demonstrate the excellent red-teaming performance of our human-readable, filter-resistant adversarial prompts, as well as superior zero-shot transferability which enables instant adaptation to unseen prompts and exposes critical vulnerabilities even in commercial APIs (e.g., Leonardo.Ai.).
title AutoPrompt: Automated Red-Teaming of Text-to-Image Models via LLM-Driven Adversarial Prompts
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.24034