mlsanityData to 7 Oct 2026

R.A.C.E.: Robust Adversarial Concept Erasure for Secure Text-to-Image Diffusion Model

arXiv 2405.16341 ↗First posted 25 May 2024cs.CVcited by 56

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

Abstract (arXiv)

In the evolving landscape of text-to-image (T2I) diffusion models, the remarkable capability to generate high-quality images from textual descriptions faces challenges with the potential misuse of reproducing sensitive content. To address this critical issue, we introduce Robust Adversarial Concept Erase (RACE), a novel approach designed to mitigate these risks by enhancing the robustness of concept erasure method for T2I models. RACE utilizes a sophisticated adversarial training framework to identify and mitigate adversarial text embeddings, significantly reducing the Attack Success Rate (ASR). Impressively, RACE achieves a 30 percentage point reduction in ASR for the ``nudity'' concept against the leading white-box attack method. Our extensive evaluations demonstrate RACE's effectiveness in defending against both white-box and black-box attacks, marking a significant advancement in protecting T2I diffusion models from generating inappropriate or misleading imagery. This work underlines the essential need for proactive defense measures in adapting to the rapidly advancing field of adversarial challenges. Our code is publicly available: urlhttps://github.com/chkimmmmm/R.A.C.E.

Key facts. Cited by 56 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 66 of its 74 references that are in the map. First posted on arXiv 25 May 2024.

Builds on

66 of 74 references.

Closest ancestors · 18

Same direction, newest first

  1. 2023One-Dimensional Adapter to Rule Them All: Concepts, Diffusion Models and Erasing Applications26 Dec
  2. 2023All but One: Surgical Concept Erasing with Model Preservation in Text-to-Image Diffusion Models20 Dec
  3. 2023Receler: Reliable Concept Erasing of Text-to-Image Diffusion Models via Lightweight Erasers29 Nov
  4. 2023MMA-Diffusion: MultiModal Attack on Diffusion Models29 Nov
  5. 2023To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images ... For Now18 Oct
  6. 2023Ring-A-Bell! How Reliable are Concept Removal Methods for Diffusion Models?16 Oct
  7. 2023Prompting4Debugging: Red-Teaming Text-to-Image Diffusion Models by Finding Problematic Prompts12 Sep
  8. 2023Backdooring Textual Inversion for Concept Censorship21 Aug
  9. 2023Towards Safe Self-Distillation of Internet-Scale Text-to-Image Diffusion Models12 Jul
  10. 2023WOUAF: Weight Modulation for User Attribution and Fingerprinting in Text-to-Image Diffusion Models7 Jun
  11. 2023Tree-Ring Watermarks: Fingerprints for Diffusion Images that are Invisible and Robust31 May
  12. 2023Forget-Me-Not: Learning to Forget in Text-to-Image Diffusion Models30 Mar
6 more
  1. 2023The Stable Signature: Rooting Watermarks in Latent Diffusion Models27 Mar
  2. 2023Ablating Concepts in Text-to-Image Diffusion Models23 Mar
  3. 2023Erasing Concepts from Diffusion Models13 Mar
  4. 2023Extracting Training Data from Diffusion Models30 Jan
  5. 2022Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models9 Nov
  6. 2022Red-Teaming the Stable Diffusion Safety Filter3 Oct

Other references · 48

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  3. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  4. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  5. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  6. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  7. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  8. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  9. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,539
  10. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  11. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,322
  12. 2022Classifier-Free Diffusion Guidancecited by 5,717
36 more
  1. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,131
  2. 2021GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Modelscited by 3,595
  3. 2019Analyzing and Improving the Image Quality of StyleGANcited by 3,580
  4. 2022DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generationcited by 3,286
  5. 2020Taming Transformers for High-Resolution Image Synthesiscited by 3,229
  6. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  7. 2021CLIPScore: A Reference-free Evaluation Metric for Image Captioningcited by 2,388
  8. 2022An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversioncited by 2,231
  9. 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,510
  10. 2022Multi-Concept Customization of Text-to-Image Diffusioncited by 1,099
  11. 2018Robustness May Be at Odds with Accuracycited by 873
  12. 2017Robust Adversarial Reinforcement Learningcited by 439
  13. 2022Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Modelscited by 418
  14. 2023ELITE: Encoding Visual Concepts into Textual Embeddings for Customized Text-to-Image Generationcited by 414
  15. 2020Visual Transformers: Token-based Image Representation and Processing for Computer Visioncited by 333
  16. 2023Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discoverycited by 329
  17. 2023Your Diffusion Model is Secretly a Zero-Shot Classifiercited by 301
  18. 2023Understanding and Mitigating Copying in Diffusion Modelscited by 205
  19. 2020Artificial Fingerprinting for Generative Models: Rooting Deepfake Attribution in Training Datacited by 200
  20. 2020fastai: A Layered API for Deep Learningcited by 183
  21. 2015Large-scale Classification of Fine-Art Paintings: Learning The Right Metric on The Right Featurecited by 181
  22. 2023Selective Amnesia: A Continual Learning Approach to Forgetting in Deep Generative Modelscited by 175
  23. 2019Robust Invisible Video Watermarking with Attentioncited by 144
  24. 2021Feature Importance-aware Transferable Adversarial Attackscited by 125
  25. 2020Responsible Disclosure of Generative Models Using Scalable Fingerprintingcited by 90
  26. 2023Circumventing Concept Erasure Methods For Text-to-Image Generative Modelscited by 81
  27. 2021Towards Improving Adversarial Training of NLP Modelscited by 79
  28. 2024Get What You Want, Not What You Don't: Image Content Suppression for Text-to-Image Diffusion Modelscited by 62
  29. 2021Meta Gradient Adversarial Attackcited by 55
  30. 2024λ-ECLIPSE: Multi-Concept Personalized Text-to-Image Diffusion Models by Leveraging CLIP Latent Spacecited by 51
  31. 2023ECLIPSE: A Resource-Efficient Text-to-Image Prior for Image Generationscited by 24
  32. 2023Attributing Image Generative Models using Latent Fingerprintscited by 20
  33. 2023IMMA: Immunizing text-to-image Models against Malicious Adaptationcited by 17
  34. 2020Decentralized Attribution of Generative Modelscited by 16
  35. 2023ORES: Open-vocabulary Responsible Visual Synthesiscited by 14
  36. 2022Analysis and Extensions of Adversarial Training for Video Classificationcited by 11

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 13; 2025: 25; 2026: 18.

56 papers.

Sort:

By area:

  1. 2024SAFREE: Training-Free and Adaptive Guard for Safe Text-to-Image And Video Generationcited by 125
  2. 2025Set You Straight: Auto-Steering Denoising Trajectories to Sidestep Unwanted Conceptscited by 61
  3. 2024TripletCLIP: Improving Compositional Reasoning of CLIP via Synthetic Vision-Language Negativescited by 57
  4. 2024STEREO: A Two-Stage Framework for Adversarially Robust Concept Erasing from Text-to-Image Diffusion Modelscited by 39
  5. 2024Meta-Unlearning on Diffusion Models: Preventing Relearning Unlearned Conceptscited by 39
  6. 2024Steering Rectified Flow Models in the Vector Field for Controlled Image Generationcited by 38
  7. 2024Adversarial Attacks and Defenses on Text-to-Image Diffusion Models: A Surveycited by 34
  8. 2025SPEED: Scalable, Precise, and Efficient Concept Erasure for Diffusion Modelscited by 33
  9. 2024Attacks and Defenses for Generative Diffusion Models: A Comprehensive Surveycited by 32
  10. 2025CURE: Concept Unlearning via Orthogonal Representation Editing in Diffusion Modelscited by 24
  11. 2025Training-Free Safe Denoisers for Safe Use of Diffusion Modelscited by 19
  12. 2025One Image is Worth a Thousand Words: A Usability Preservable Text-Image Collaborative Erasing Frameworkcited by 14
38 more
  1. 2024Backdooring Bias (B2) into Stable Diffusion Modelscited by 12
  2. 2024Unveiling Concept Attribution in Diffusion Modelscited by 11
  3. 2025Concept Corrector: Erase concepts on the fly for text-to-image diffusion modelscited by 11
  4. 2025Erasing Concepts, Steering Generations: A Comprehensive Survey of Concept Suppressioncited by 10
  5. 2025A Comprehensive Survey on Concept Erasure in Text-to-Image Diffusion Modelscited by 9
  6. 2025Erased or Dormant? Rethinking Concept Erasure Through Reversibilitycited by 9
  7. 2025Towards Resilient Safety-driven Unlearning for Diffusion Models against Downstream Fine-tuningcited by 8
  8. 2025SuMa: A Subspace Mapping Approach for Robust and Effective Concept Erasure in Text-to-Image Diffusion Modelscited by 8
  9. 2025Bridging Symmetry and Robustness: On the Role of Equivariance in Enhancing Adversarial Robustnesscited by 8
  10. 2025EraseFlow: Learning Concept Erasure Policies via GFlowNet-Driven Alignmentcited by 8
  11. 2025CE-SDWV: Effective and Efficient Concept Erasure for Text-to-Image Diffusion Models via a Semantic-Driven Word Vocabularycited by 7
  12. 2025TarPro: Targeted Protection against Malicious Image Editingcited by 7
  13. 2025Semantic Surgery: Zero-Shot Concept Erasure in Diffusion Modelscited by 6
  14. 2024Safe Text-to-Image Generation: Simply Sanitize the Prompt Embeddingcited by 4
  15. 2025EAR: Erasing Concepts from Unified Autoregressive Modelscited by 4
  16. 2025ACE: Attentional Concept Erasure in Diffusion Modelscited by 3
  17. 2025CGCE: Classifier-Guided Concept Erasure in Generative Modelscited by 3
  18. 2026UnHype: CLIP-Guided Hypernetworks for Dynamic LoRA Unlearningcited by 3
  19. 2026To Erase, or Not to Erase: Robust Training-Free Concept Erasure with Preservation aware Adaptive Ranked Subspace Expansioncited by 3
  20. 2025Rethinking the Vulnerability of Concept Erasure and a New Methodcited by 2
  21. 2026ActErase: A Training-Free Paradigm for Precise Concept Erasure via Activation Redirectioncited by 2
  22. 2026Orthogonal Concept Erasure for Diffusion Modelscited by 2
  23. 2026GEM: Geometric Erasure by Contrastive Velocity Matching in Rectified Flowscited by 2
  24. 2024Red-Teaming Text-to-Image Models via In-Context Experience Replay and Semantic-Preserving Prompt Rewritingcited by 1
  25. 2025SAGE: Exploring the Boundaries of Unsafe Concept Domain with Semantic-Augment Erasingcited by 1
  26. 2025ZIUM: Zero-Shot Intent-Aware Adversarial Attack on Unlearned Modelscited by 1
  27. 2025Beyond Fixed Anchors: Precisely Erasing Concepts with Sibling Exclusive Counterpartscited by 1
  28. 2025Rethinking Robust Adversarial Concept Erasure in Diffusion Modelscited by 1
  29. 2026Prototype-Guided Concept Erasure in Diffusion Modelscited by 1
  30. 2026Erasure or Erosion? Evaluating Compositional Degradation in Unlearned Text-To-Image Diffusion Modelscited by 1
  31. 2026Beyond Text Prompts: Precise Concept Erasure through Text-Image Collaborationcited by 1
  32. 2026FlowErase-RL: Rethinking Concept Erasure as Reward Optimization in Flow Matching Modelscited by 1
  33. 2024Dark Miner: Defend against undesirable generation for text-to-image diffusion modelscited by 0
  34. 2024FameBias: Embedding Manipulation Bias Attack in Text-to-Image Modelscited by 0
  35. 2025Bi-Erasing: A Bidirectional Framework for Concept Removal in Diffusion Modelscited by 0
  36. 2026Consistency-Preserving Concept Erasure via Unsafe-Safe Pairing and Directional Fisher-weighted Adaptationcited by 0
  37. 2026When Safe Concepts Become Unsafe: Multi-Concept Compositional Vulnerabilities in Text-to-Image Modelscited by 0
  38. 2026CoreUnlearn: Rethinking Concept Unlearning through Disentangled Component-Level Erasure in Text-guided Diffusion Modelscited by 0

50 most cited of 56 All 56 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024Erasing Undesirable Concepts in Diffusion Models with Adversarial Preservation
  2. 2024Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models
  3. 2026DSS: Dynamic Semantic Steering for Robust Concept Erasure in Diffusion Models
  4. 2025FADE: Adversarial Concept Erasure in Flow Models
  5. 2025Robust Concept Erasure in Diffusion Models: A Theoretical Perspective on Security and Robustness
  6. 2024Probing Unlearned Diffusion Models: A Transferable Adversarial Attack Perspective
  7. 2026Certifying Concept Unlearning in Text-to-Image Diffusion Models
  8. 2024Pruning for Robust Concept Erasing in Diffusion Models