mlsanityData to 7 Oct 2026

Erasing Concepts from Diffusion Models

arXiv 2303.07345 ↗First posted 13 Mar 2023cs.CVcited by 530

On the map: Diffusion models → Robustness, privacy and adversarial aspects

Abstract (arXiv)

Motivated by recent advancements in text-to-image diffusion, we study erasure of specific concepts from the model's weights. While Stable Diffusion has shown promise in producing explicit or realistic artwork, it has raised concerns regarding its potential for misuse. We propose a fine-tuning method that can erase a visual concept from a pre-trained diffusion model, given only the name of the style and using negative guidance as a teacher. We benchmark our method against previous approaches that remove sexually explicit content and demonstrate its effectiveness, performing on par with Safe Latent Diffusion and censored training. To evaluate artistic style removal, we conduct experiments erasing five modern artists from the network and conduct a user study to assess the human perception of the removed styles. Unlike previous methods, our approach can remove concepts from a diffusion model permanently rather than modifying the output at the inference time, so it cannot be circumvented even if a user has access to model weights. Our code, data, and results are available at https://erasing.baulab.info/

Key facts. Cited by 530 papers in the map (data through 7 Oct 2026). Placed in Robustness, privacy and adversarial aspects, one of 14 directions in Diffusion models. Builds on 32 of its 49 references that are in the map. First posted on arXiv 13 Mar 2023.

Builds on

32 of 49 references.

Closest ancestors · 3

Same direction, newest first

  1. 2023Raising the Cost of Malicious AI-Powered Image Editing13 Feb
  2. 2022Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models9 Nov
  3. 2022Red-Teaming the Stable Diffusion Safety Filter3 Oct

Other references · 29

  1. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  3. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  4. 2022Classifier-Free Diffusion Guidancecited by 5,717
  5. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,131
  6. 2021GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Modelscited by 3,595
  7. 2022DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generationcited by 3,286
  8. 2022Locating and Editing Factual Associations in GPTcited by 2,609
  9. 2022An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversioncited by 2,231
  10. 2022Multi-Concept Customization of Text-to-Image Diffusioncited by 1,099
  11. 2020Improved Techniques for Training Score-Based Generative Modelscited by 1,036
  12. 2019Machine Unlearningcited by 1,020
17 more
  1. 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 847
  2. 2022Quantifying Memorization Across Neural Language Modelscited by 755
  3. 2022Compositional Visual Generation with Composable Diffusion Modelscited by 619
  4. 2019Eternal Sunshine of the Spotless Net: Selective Forgetting in Deep Networkscited by 599
  5. 2021Knowledge Neurons in Pretrained Transformerscited by 590
  6. 2021Editing Factual Knowledge in Language Modelscited by 548
  7. 2021Fast Model Editing at Scalecited by 491
  8. 2022Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Modelscited by 418
  9. 2021Remember What You Want to Forget: Algorithms for Machine Unlearningcited by 360
  10. 2021Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLPcited by 336
  11. 2023Glaze: Protecting Artists from Style Mimicry by Text-to-Image Modelscited by 200
  12. 2020fastai: A Layered API for Deep Learningcited by 183
  13. 2021StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generatorscited by 164
  14. 2020Rewriting a Deep Generative Modelcited by 105
  15. 2021Unsupervised Learning of Compositional Energy Conceptscited by 71
  16. 2021Sketch Your Own GANcited by 37
  17. 2022Rewriting Geometric Rules of a GANcited by 25

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 54; 2024: 145; 2025: 164; 2026: 167.

530 papers.

Sort:

By area:

  1. 2023Unified Concept Editing in Diffusion Modelscited by 377
  2. 2023SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generationcited by 376
  3. 2023Editing Large Language Models: Problems, Methods, and Opportunitiescited by 367
  4. 2023Ablating Concepts in Text-to-Image Diffusion Modelscited by 312
  5. 2024MACE: Mass Concept Erasure in Diffusion Modelscited by 298
  6. 2023Forget-Me-Not: Learning to Forget in Text-to-Image Diffusion Modelscited by 297
  7. 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Modelscited by 251
  8. 2024Rethinking Machine Unlearning for Large Language Modelscited by 247
  9. 2023Ring-A-Bell! How Reliable are Concept Removal Methods for Diffusion Models?cited by 239
  10. 2023Model Sparsity Can Simplify Machine Unlearningcited by 206
  11. 2023To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images ... For Nowcited by 204
  12. 2023MMA-Diffusion: MultiModal Attack on Diffusion Modelscited by 184
38 more
  1. 2023Selective Amnesia: A Continual Learning Approach to Forgetting in Deep Generative Modelscited by 175
  2. 2023One-Dimensional Adapter to Rule Them All: Concepts, Diffusion Models and Erasing Applicationscited by 164
  3. 2024Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Modelscited by 164
  4. 2023Prompting4Debugging: Red-Teaming Text-to-Image Diffusion Models by Finding Problematic Promptscited by 161
  5. 2023Holistic Evaluation of Text-To-Image Modelscited by 160
  6. 2024A Comprehensive Study of Knowledge Editing for Large Language Modelscited by 157
  7. 2023Can Sensitive Information Be Deleted From LLMs? Objectives for Defending Against Extraction Attackscited by 156
  8. 2023Concept Sliders: LoRA Adaptors for Precise Control in Diffusion Modelscited by 148
  9. 2023LEDITS++: Limitless Image Editing using Text-to-Image Modelscited by 147
  10. 2024SAFREE: Training-Free and Adaptive Guard for Safe Text-to-Image And Video Generationcited by 125
  11. 2024Reliable and Efficient Concept Erasure of Text-to-Image Diffusion Modelscited by 123
  12. 2023Receler: Reliable Concept Erasing of Text-to-Image Diffusion Models via Lightweight Eraserscited by 118
  13. 2024Robust Watermarking Using Generative Priors Against Image Editing: From Benchmarking to Advancescited by 105
  14. 2024Model Editing Harms General Abilities of Large Language Models: Regularization to the Rescuecited by 101
  15. 2024EraseAnything: Enabling Concept Erasure in Rectified Flow Transformerscited by 93
  16. 2023Circumventing Concept Erasure Methods For Text-to-Image Generative Modelscited by 81
  17. 2023Dynamic Prompt Learning: Addressing Cross-Attention Leakage for Text-Based Image Editingcited by 76
  18. 2025SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoderscited by 75
  19. 2024Adaptive Activation Steering: A Tuning-Free LLM Truthfulness Improvement Method for Diverse Hallucinations Categoriescited by 74
  20. 2023Self-Discovering Interpretable Diffusion Latent Directions for Responsible Text-to-Image Generationcited by 69
  21. 2024Jailbreaking Prompt Attack: A Controllable Adversarial Attack against Diffusion Modelscited by 68
  22. 2024Latent Guard: a Safety Framework for Text-to-image Generationcited by 68
  23. 2024T2VSafetyBench: Evaluating the Safety of Text-to-Video Generative Modelscited by 66
  24. 2023Toward effective protection against diffusion based mimicry through score distillationcited by 64
  25. 2024SafeGen: Mitigating Sexually Explicit Content Generation in Text-to-Image Modelscited by 64
  26. 2023Inserting Anybody in Diffusion Models via Celeb Basiscited by 63
  27. 2024Get What You Want, Not What You Don't: Image Content Suppression for Text-to-Image Diffusion Modelscited by 62
  28. 2024GuardT2I: Defending Text-to-Image Models from Adversarial Promptscited by 61
  29. 2025Set You Straight: Auto-Steering Denoising Trajectories to Sidestep Unwanted Conceptscited by 61
  30. 2024MyVLM: Personalizing VLMs for User-Specific Queriescited by 58
  31. 2024R.A.C.E.: Robust Adversarial Concept Erasure for Secure Text-to-Image Diffusion Modelcited by 56
  32. 2024Erasing Undesirable Influence in Diffusion Modelscited by 52
  33. 2024UnsafeBench: Benchmarking Image Safety Classifiers on Real-World and AI-Generated Imagescited by 52
  34. 2024ConceptPrune: Concept Editing in Diffusion Models via Skilled Neuron Pruningcited by 51
  35. 2024Scissorhands: Scrub Data Influence via Connection Sensitivity in Networkscited by 50
  36. 2024Direct Unlearning Optimization for Robust and Safe Text-to-Image Modelscited by 50
  37. 2024Erasing Undesirable Concepts in Diffusion Models with Adversarial Preservationcited by 48
  38. 2024Machine Unlearning for Image-to-Image Generative Modelscited by 47

50 most cited of 530 All 530 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2023All but One: Surgical Concept Erasing with Model Preservation in Text-to-Image Diffusion Models
  2. 2026Erasing Thousands of Concepts: Towards Scalable and Practical Concept Erasure for Text-to-Image Diffusion Models
  3. 2025CASteer: Cross-Attention Steering for Controllable Concept Erasure
  4. 2025Less is More: Masking Elements in Image Condition Features Avoids Content Leakages in Style Transfer Diffusion Models
  5. 2026The Silent Brush: Evaluating Artistic Style Leakage in AI Art Generation
  6. 2026Compositional Concept Erasure in Text-to-Image Diffusion Models via Hierarchically Grounded Semantic Surgery
  7. 2024Unlearning Concepts in Diffusion Model via Concept Domain Correction and Concept Preserving Gradient
  8. 2025The Cow of Rembrandt - Analyzing Artistic Prompt Interpretation in Text-to-Image Models