MACE: Mass Concept Erasure in Diffusion Models
This addresses misuse concerns in AI-generated imagery, offering a scalable solution for content moderation, though it is incremental as it builds on existing concept erasure techniques.
The paper tackles the problem of preventing text-to-image diffusion models from generating harmful or misleading content by introducing MACE, a finetuning framework for mass concept erasure that scales up to 100 concepts and outperforms prior methods across tasks like object, celebrity, explicit content, and artistic style erasure.
The rapid expansion of large-scale text-to-image diffusion models has raised growing concerns regarding their potential misuse in creating harmful or misleading content. In this paper, we introduce MACE, a finetuning framework for the task of mass concept erasure. This task aims to prevent models from generating images that embody unwanted concepts when prompted. Existing concept erasure methods are typically restricted to handling fewer than five concepts simultaneously and struggle to find a balance between erasing concept synonyms (generality) and maintaining unrelated concepts (specificity). In contrast, MACE differs by successfully scaling the erasure scope up to 100 concepts and by achieving an effective balance between generality and specificity. This is achieved by leveraging closed-form cross-attention refinement along with LoRA finetuning, collectively eliminating the information of undesirable concepts. Furthermore, MACE integrates multiple LoRAs without mutual interference. We conduct extensive evaluations of MACE against prior methods across four different tasks: object erasure, celebrity erasure, explicit content erasure, and artistic style erasure. Our results reveal that MACE surpasses prior methods in all evaluated tasks. Code is available at https://github.com/Shilin-LU/MACE.