Concept Removal for Frontier Image Generative Models
For developers and deployers of large-scale image generative models, this provides a practical and persistent method to remove undesirable visual concepts without degrading output quality.
The paper introduces a concept removal method for frontier image generative models (e.g., SD3.5, Flux, Infinity) that replaces the internal bottleneck layer with a transcoder to selectively disable concept-specific signals. The method achieves state-of-the-art concept removal performance while maintaining visual quality and robustness against adversarial prompts.
Image generative models are trained on massive, largely uncurated internet-scale datasets that contain undesirable visual concepts. Efficiently removing such concepts from the model generations without degrading the quality of output images remains challenging. We introduce a novel concept removal method for frontier diffusion and image autoregressive models, such as SD3.5, Flux, and Infinity. Our intervention replaces the internal bottleneck layer present in all these modern models with a transcoder that is trained to replicate the original layer while structuring it into distinct activation features. This in-place substitution creates an integrated filter through which concept-specific signals can be selectively disabled while preserving the rest of the model's behavior. Since the intervention modifies the model backbone rather than attaching an external component, it remains persistent under white-box access. Empirically, the approach achieves state-of-the-art concept removal performance across modern diffusion and autoregressive models, maintains visual generation quality, provides robustness against adversarial prompts, and supports sequential removal of diverse concepts. This positions our method as a practical approach for concept removal in frontier image generative models.