Search papers, labs, and topics across Lattice.
This paper addresses the critical challenge of ensuring safety and policy compliance in text-to-image diffusion models by proposing a novel approach that leverages intermediate clean image estimates during the generation process. The method utilizes a sparse margin objective to detect prohibited content and intervenes by optimizing a structured low-rank residual in the text-conditioning space, allowing for effective weight-preserving detection. Experimental results on Stable Diffusion v1.4 and v3.5 show that this approach significantly outperforms existing methods in terms of suppression, fidelity, and preservation, providing a scalable solution for safe generative deployment.
A novel test-time scaling method achieves superior safety in text-to-image generation without compromising inference speed or general capabilities.
Ensuring safety and policy compliance in text-to-image diffusion models remains a critical challenge, as benign or adversarial prompts can often elicit prohibited content, e.g. nudity and protected intellectual property. While training-based unlearning methods are effective, they are computationally expensive and prone to catastrophic interference with general capabilities. Conversely, existing test-time defenses are primarily prompt-centric, relying on modifying textual descriptions only, and overlook the visual signals for detection. In this paper, we propose to leverage the intermediate clean image estimated during the generation process and employ a sparse margin objective to detect prohibited concepts. When a violation is detected, we immediately intervene by optimizing a structured low-rank residual in the text-conditioning space via truncated backpropagation. This design allows weight-preserving detection, keeps non-violating inference latency nearly unchanged as the maximum budget increases, and offers flexibility in safety performance via test-time scaling. Extensive experiments on Stable Diffusion v1.4 and v3.5 across nudity removal, IP protection, and style erasure demonstrate superior performance across suppression, fidelity and preservation compared to prior weight-preserving baselines, providing a scalable and flexible solution for safe generative deployment.
