Modular Energy Steering for Safe Text-to-Image Generation with Foundation Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Tan, Yaoteng, Cai, Zikui, Asif, M. Salman
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914440681619456
author Tan, Yaoteng
Cai, Zikui
Asif, M. Salman
author_facet Tan, Yaoteng
Cai, Zikui
Asif, M. Salman
contents Controlling the behavior of text-to-image generative models is critical for safe and practical deployment. Existing safety approaches typically rely on model fine-tuning or curated datasets, which can degrade generation quality or limit scalability. We propose an inference-time steering framework that leverages gradient feedback from frozen pretrained foundation models to guide the generation process without modifying the underlying generator. Our key observation is that vision-language foundation models encode rich semantic representations that can be repurposed as off-the-shelf supervisory signals during generation. By injecting such feedback through clean latent estimates at each sampling step, our method formulates safety steering as an energy-based sampling problem. This design enables modular, training-free safety control that is compatible with both diffusion and flow-matching models and can generalize across diverse visual concepts. Experiments demonstrate state-of-the-art robustness against NSFW red-teaming benchmarks and effective multi-target steering, while preserving high generation quality on benign non-targeted prompts. Our framework provides a principled approach for utilizing foundation models as semantic energy estimators, enabling reliable and scalable safety control for text-to-image generation.
format Preprint
id arxiv_https___arxiv_org_abs_2604_02265
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Modular Energy Steering for Safe Text-to-Image Generation with Foundation Models
Tan, Yaoteng
Cai, Zikui
Asif, M. Salman
Computer Vision and Pattern Recognition
Controlling the behavior of text-to-image generative models is critical for safe and practical deployment. Existing safety approaches typically rely on model fine-tuning or curated datasets, which can degrade generation quality or limit scalability. We propose an inference-time steering framework that leverages gradient feedback from frozen pretrained foundation models to guide the generation process without modifying the underlying generator. Our key observation is that vision-language foundation models encode rich semantic representations that can be repurposed as off-the-shelf supervisory signals during generation. By injecting such feedback through clean latent estimates at each sampling step, our method formulates safety steering as an energy-based sampling problem. This design enables modular, training-free safety control that is compatible with both diffusion and flow-matching models and can generalize across diverse visual concepts. Experiments demonstrate state-of-the-art robustness against NSFW red-teaming benchmarks and effective multi-target steering, while preserving high generation quality on benign non-targeted prompts. Our framework provides a principled approach for utilizing foundation models as semantic energy estimators, enabling reliable and scalable safety control for text-to-image generation.
title Modular Energy Steering for Safe Text-to-Image Generation with Foundation Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.02265