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

ArXi:2604.02265v1 Announce Type: new 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.