AI RESEARCH
When Safety Collides: Resolving Multi-Category Harmful Conflicts in Text-to-Image Diffusion via Adaptive Safety Guidance
arXiv CS.CV
•
ArXi:2602.20880v4 Announce Type: replace Text-to-Image (T2I) diffusion models have nstrated significant advancements in generating high-quality images, while raising potential safety concerns regarding harmful content generation. Safety-guidance-based methods have been proposed to mitigate harmful outputs by steering generation away from harmful zones, where the zones are averaged across multiple harmful categories based on predefined keywords.