# Consistency-Preserving Concept Erasure via Unsafe-Safe Pairing and Directional Fisher-weighted Adaptation

arXiv: https://arxiv.org/abs/2602.05339 · HTML: https://mlsanity.com/paper/2602.05339/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Robustness, privacy and adversarial aspects, one of 14 directions in Diffusion models.
- Builds on 39 of its 52 references that are in the map.
- First posted on arXiv 5 Feb 2026.

## Placed in

- Diffusion models / [Robustness, privacy and adversarial aspects](https://mlsanity.com/diffusion/robustness-privacy-and-adversarial-aspects/)
- LLM safety and security / [Vision-language and text-to-image safety](https://mlsanity.com/safety/vision-language-and-text-to-image-safety/)

## Builds on (most cited first, 30 of 39)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2403.03206/) (2024)
- [LAION-5B: An open large-scale dataset for training next generation image-text models](https://mlsanity.com/paper/2210.08402/) (2022)
- [CLIPScore: A Reference-free Evaluation Metric for Image Captioning](https://mlsanity.com/paper/2104.08718/) (2021)
- [Conditional Generative Adversarial Nets](https://mlsanity.com/paper/1411.1784/) (2014)
- [IP-Adapter: Text Compatible Image Prompt Adapter for Text-to-Image Diffusion Models](https://mlsanity.com/paper/2308.06721/) (2023)
- [Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction](https://mlsanity.com/paper/2404.02905/) (2024)
- [Erasing Concepts from Diffusion Models](https://mlsanity.com/paper/2303.07345/) (2023)
- [Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models](https://mlsanity.com/paper/2211.05105/) (2022)
- [SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation](https://mlsanity.com/paper/2310.12508/) (2023)
- [Ablating Concepts in Text-to-Image Diffusion Models](https://mlsanity.com/paper/2303.13516/) (2023)
- [A Survey of Machine Unlearning](https://mlsanity.com/paper/2209.02299/) (2022)
- [Forget-Me-Not: Learning to Forget in Text-to-Image Diffusion Models](https://mlsanity.com/paper/2303.17591/) (2023)
- [Ring-A-Bell! How Reliable are Concept Removal Methods for Diffusion Models?](https://mlsanity.com/paper/2310.10012/) (2023)
- [To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images ... For Now](https://mlsanity.com/paper/2310.11868/) (2023)
- [Understanding and Mitigating Copying in Diffusion Models](https://mlsanity.com/paper/2305.20086/) (2023)
- [In-Context Edit: Enabling Instructional Image Editing with In-Context Generation in Large Scale Diffusion Transformer](https://mlsanity.com/paper/2504.20690/) (2025)
- [MMA-Diffusion: MultiModal Attack on Diffusion Models](https://mlsanity.com/paper/2311.17516/) (2023)
- [Reliable and Efficient Concept Erasure of Text-to-Image Diffusion Models](https://mlsanity.com/paper/2407.12383/) (2024)
- [SAFREE: Training-Free and Adaptive Guard for Safe Text-to-Image And Video Generation](https://mlsanity.com/paper/2410.12761/) (2024)
- [Prometheus-Vision: Vision-Language Model as a Judge for Fine-Grained Evaluation](https://mlsanity.com/paper/2401.06591/) (2024)
- [Unveiling and Mitigating Memorization in Text-to-image Diffusion Models through Cross Attention](https://mlsanity.com/paper/2403.11052/) (2024)
- [R.A.C.E.: Robust Adversarial Concept Erasure for Secure Text-to-Image Diffusion Model](https://mlsanity.com/paper/2405.16341/) (2024)
- [Into the LAIONs Den: Investigating Hate in Multimodal Datasets](https://mlsanity.com/paper/2311.03449/) (2023)
- [T2ISafety: Benchmark for Assessing Fairness, Toxicity, and Privacy in Image Generation](https://mlsanity.com/paper/2501.12612/) (2025)
- [STEREO: A Two-Stage Framework for Adversarially Robust Concept Erasing from Text-to-Image Diffusion Models](https://mlsanity.com/paper/2408.16807/) (2024)
- [OmniPaint: Mastering Object-Oriented Editing via Disentangled Insertion-Removal Inpainting](https://mlsanity.com/paper/2503.08677/) (2025)

## Abstract (arXiv)

With the increasing versatility of text-to-image diffusion models, the ability to selectively erase undesirable concepts (e.g., harmful content) has become indispensable. However, existing concept erasure approaches primarily focus on removing unsafe concepts without providing guidance toward corresponding safe alternatives, which often leads to failure in preserving the structural and semantic consistency between the original and erased generations. In this paper, we propose a novel framework, PAIRed Erasing (PAIR), which reframes concept erasure from simple removal to consistency-preserving semantic realignment using unsafe-safe pairs. We first generate safe counterparts from unsafe inputs while preserving structural and semantic fidelity, forming paired unsafe-safe multimodal data. Leveraging these pairs, we introduce two key components: (1) Paired Semantic Realignment, a guided objective that uses unsafe-safe pairs to explicitly map target concepts to semantically aligned safe anchors; and (2) Fisher-weighted Initialization for DoRA, which initializes parameter-efficient low-rank adaptation matrices using unsafe-safe pairs, encouraging the generation of safe alternatives while selectively suppressing unsafe concepts. Together, these components enable fine-grained erasure that removes only the targeted concepts while maintaining overall semantic consistency. Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art baselines, achieving effective concept erasure while preserving structural integrity, semantic coherence, and generation quality.