analiza aiqbench
This research addresses critical security vulnerabilities in text-to-image diffusion models by introducing Dynamic Semantic Steering (DSS), a training-free defense mechanism designed for robust concept erasure. By leveraging geometry-aware formulations that model local semantic neighborhoods, the method effectively suppresses sensitive content while minimizing semantic drift and preserving generation fidelity under adversarial prompts. The approach offers a practical solution to the trade-off between flexibility and cost inherent in current AI safety strategies.
DSS: Dynamic Semantic Steering for Robust Concept Erasure in Diffusion Models
arXiv:2604.16483v2 Announce Type: replace-cross Abstract: Text-to-image (T2I) diffusion models have introduced new security risks, as adversaries can exploit flexible text prompts to induce the generation of sensitive or policy-violating content (e.g., NSFW or copyrighted concepts). Concept erasure has emerged as a promising defense, aiming to suppress targeted semantics while preserving benign generation. However, existing approaches face a fundamental trade-off: training-based methods are costly and inflexible to emerging threats, while inference-time interventions often rely on unconstrained feature manipulation, leading to over-correction, semantic drift, and degraded utility. More critically, they lack an explicit understanding of local semantic structure, resulting in unstable behavior under adversarial or compositional prompts. In this paper, we propose Dynamic Semantic Steering (DSS), a training-free, inference-time defense framework for robust and controllable concept erasure under prompt-level adversarial settings. DSS introduces a geometry-aware formulation that explicitly models local semantic neighborhoods and constrains feature updates within this structure, which is key to achieving precise suppression without sacrificing benign semantics. Specifically, DSS (i) automatically identifies benign semantic anchors via density-based boundary modeling, and (ii) performs context-aware, constrained feature correction using cross-attention signals with a closed-form solution. Extensive experiments demonstrate that DSS achieves strong and consistent suppression across diverse concept categories and adversarial prompts, reaching an average erasure rate of 91.0\%, outperforming prior defenses (18.6\%--85.9\%). At the same time, DSS substantially reduces semantic drift and preserves generation fidelity.
Abstract:Text-to-image (T2I) diffusion models have introduced new security risks, as adversaries can exploit flexible text prompts to induce the generation of sensitive or policy-violating content (e.g., NSFW or copyrighted concepts). Concept erasure has emerged as a promising defense, aiming to suppress targeted semantics while preserving benign generation. However, existing approaches face a fundamental trade-off: training-based methods are costly and inflexible to emerging threats, while inference-time interventions often rely on unconstrained feature manipulation, leading to over-correction, semantic drift, and degraded utility. More critically, they lack an explicit understanding of local semantic structure, resulting in unstable behavior under adversarial or compositional prompts. In this paper, we propose Dynamic Semantic Steering (DSS), a training-free, inference-time defense framework for robust and controllable concept erasure under prompt-level adversarial settings. DSS introduces a geometry-aware formulation that explicitly models local semantic neighborhoods and constrains feature updates within this structure, which is key to achieving precise suppression without sacrificing benign semantics. Specifically, DSS (i) automatically identifies benign semantic anchors via density-based boundary modeling, and (ii) performs context-aware, constrained feature correction using cross-attention signals with a closed-form solution. Extensive experiments demonstrate that DSS achieves strong and consistent suppression across diverse concept categories and adversarial prompts, reaching an average erasure rate of 91.0\%, outperforming prior defenses (18.6\%--85.9\%). At the same time, DSS substantially reduces semantic drift and preserves generation fidelity.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2604.16483 [cs.CV]
(or arXiv:2604.16483v2 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2604.16483 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 318 cuvinte
Subiecte conexe
To Erase, or Not to Erase: Robust Training-Free Concept Erasure with Preservation aware Adaptive Ranked Subspace Expansion
arXiv · 28 iul. 2026
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · 25 sept. 2026
EraseSAE: Surgical Concept Erasure in Text-to-Video Diffusion Models via Sparse Autoencoders
arXiv · 5 sept. 2026
CleanVideo: Adaptive Concept Erasure for Text-to-Video Diffusion Models
arXiv · 18 sept. 2026