M-ErasureBench: A Comprehensive Multimodal Evaluation Benchmark for Concept Erasure in Diffusion Models
This addresses vulnerabilities in concept erasure for diffusion models, which is crucial for preventing harmful or copyrighted content generation in real-world applications like image editing, though it is incremental as it builds on existing erasure methods.
The paper tackles the problem of concept erasure in diffusion models, which often fails for non-text inputs like embeddings and latents, and introduces M-ErasureBench, a multimodal benchmark showing existing methods have over 90% concept reproduction rate in white-box settings, and proposes IRECE, a method that reduces this rate by up to 40%.
Text-to-image diffusion models may generate harmful or copyrighted content, motivating research on concept erasure. However, existing approaches primarily focus on erasing concepts from text prompts, overlooking other input modalities that are increasingly critical in real-world applications such as image editing and personalized generation. These modalities can become attack surfaces, where erased concepts re-emerge despite defenses. To bridge this gap, we introduce M-ErasureBench, a novel multimodal evaluation framework that systematically benchmarks concept erasure methods across three input modalities: text prompts, learned embeddings, and inverted latents. For the latter two, we evaluate both white-box and black-box access, yielding five evaluation scenarios. Our analysis shows that existing methods achieve strong erasure performance against text prompts but largely fail under learned embeddings and inverted latents, with Concept Reproduction Rate (CRR) exceeding 90% in the white-box setting. To address these vulnerabilities, we propose IRECE (Inference-time Robustness Enhancement for Concept Erasure), a plug-and-play module that localizes target concepts via cross-attention and perturbs the associated latents during denoising. Experiments demonstrate that IRECE consistently restores robustness, reducing CRR by up to 40% under the most challenging white-box latent inversion scenario, while preserving visual quality. To the best of our knowledge, M-ErasureBench provides the first comprehensive benchmark of concept erasure beyond text prompts. Together with IRECE, our benchmark offers practical safeguards for building more reliable protective generative models.