The field of generative models is moving towards developing more responsible and safe models. Researchers are exploring new methods to control and suppress harmful content, while preserving the fidelity and quality of the generated images and videos. One of the key directions is the development of region-based safety control methods, which can precisely localize and suppress unsafe content. Another important area of research is the development of debiasing procedures that can mitigate biases in diffusion models. Additionally, there is a growing interest in concept erasure techniques, which aim to prevent the generation of undesired concepts while preserving the ability to synthesize high-quality images and videos. However, recent studies have also highlighted the potential side effects and limitations of these techniques. Overall, the field is moving towards developing more advanced and nuanced methods for controlling and improving the safety and quality of generative models. Noteworthy papers include: SafeCtrl, which introduces a novel detect-then-suppress paradigm for region-based safety control, and VideoEraser, which proposes a training-free framework for concept erasure in text-to-video diffusion models. DeCoDi is also notable for its debiasing procedure that can be applied to any diffusion-based image generation model.