Metraipolis › archive › dc_bc1a5df38d26
Automatic Red Teaming for Implicit Vulnerabilities of Text-to-Image Models — arXiv:2609.06094
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.06094
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T15:06:49.829Z
- SHA-256 of the archived text
b68e32e180476e1a2c79b52ab03b83cea22e6331b516ad5724bb63d2c16cbedf- Stable URL of this page
https://metraipolis.com/documento/dc_bc1a5df38d26
How to cite
Automatic Red Teaming for Implicit Vulnerabilities of Text-to-Image Models — arXiv:2609.06094 — arXiv (en), https://arxiv.org/abs/2609.06094 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bc1a5df38d26 (Metraipolis archive dc_bc1a5df38d26, sha256 b68e32e180476e1a2c79b52ab03b83cea22e6331b516ad5724bb63d2c16cbedf)
BibTeX
@misc{dcbc1a5df38d26,
title = {Automatic Red Teaming for Implicit Vulnerabilities of Text-to-Image Models — arXiv:2609.06094},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.06094},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bc1a5df38d26; sha256 b68e32e180476e1a2c79b52ab03b83cea22e6331b516ad5724bb63d2c16cbedf},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.06094. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Red-teaming Text-to-Image (T2I) models is essential for safe deployment, yet it remains particularly challenging against implicit adversarial prompts. Unlike explicit adversarial prompts that can be readily identified and blocked, implicit ones are much harder to detect: the prompts appear benign on the text surface yet still lead to inappropriate visual content. To address this, we propose Adversarial Probing for Implicit VulnErabilities (AdvPIE), a multimodal agentic framework to expose implicit vulnerabilities without requiring access to the parameters of target models. AdvPIE adopts a policy agent to generate and refine implicit adversarial prompts based on the feedback from a judge agent. To construct informative feedback, the judge agent provides modality-specific safety evaluation at both global and relative levels across iterations. To effectively leverage the feedback, we propose a novel Cumulative Adversarial Decoding strategy for the policy agent, which dynamically reweights token distributions to favor tokens that lead to more harmful images while preserving sampling diversity. Extensive experiments on standard and safety-aligned T2I models show that AdvPIE1 effectively uncovers implicit vulnerabilities, outperforming various baseline methods. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.06094 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T14:49:57.046Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).