cs.CVNov 16, 2026

ASCG: Adaptive Spatial Classifier Guidance for Surgical Concept Suppression in Diffusion Models

Authors: Younghwan Kil, Joonhyeong Park, Giung Nam, Juho Lee

Organizations: KAIST, Republic of Korea

Abstract

While text-to-image (T2I) diffusion models achieve remarkable quality, they risk producing harmful content. Existing safety mechanisms often rely on indiscriminate, global interventions that induce unintended semantic drift and degrade image quality. To address this, we propose Adaptive Spatial Classifier Guidance (ASCG), a necessity-driven framework that employs probabilistic monitoring to activate guidance selectively upon harm detection. By leveraging Grad-CAM for localized suppression, ASCG effectively removes harmful concepts while preserving superior semantic fidelity compared to conventional methods. Furthermore, we overcome the limitations of traditional metrics like NudeNet---which fails to assess semantic integrity---by introducing a VLM-based evaluation framework to provide a fine-grained, robust analysis of both safety and context preservation. Our results demonstrate that ASCG not only balances safety and fidelity but also exhibits remarkable plug-and-play synergy with existing frameworks to establish state-of-the-art performance.

Figures & tables

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Example-Based Spatial Guidance for Training-Free Concept Erasure in Diffusion Models

    Nov 16, 2026Younghwan Kil, Giung Nam, Joonhyeong Park +2Concept ErasureT2I Diffusion Models

  2. SafeCtrl: Region-Aware Safety Control for Text-to-Image Diffusion via Detect-Then-Suppress

    Apr 5, 2026Lingyun Zhang, Yu Xie, Zhongli Fang +2Safe T2I GenerationAdversarial Robustness

  3. When Safety Collides: Resolving Multi-Category Harmful Conflicts in Text-to-Image Diffusion via Adaptive Safety Guidance

    Feb 24, 2026Yongli Xiang, Ziming Hong, Zhaoqing Wang +3Safe T2I GenerationDiffusion Model Guidance