# EraseFlow: Learning Concept Erasure Policies via GFlowNet-Driven Alignment

arXiv: https://arxiv.org/abs/2511.00804 · HTML: https://mlsanity.com/paper/2511.00804/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion and multimodal model alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 57 of its 67 references that are in the map.
- First posted on arXiv 2 Nov 2025.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Diffusion and multimodal model alignment](https://mlsanity.com/alignment/diffusion-and-multimodal-model-alignment/)

## Builds on (most cited first, 30 of 57)

- [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)
- [Diffusion Models Beat GANs on Image Synthesis](https://mlsanity.com/paper/2105.05233/) (2021)
- [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://mlsanity.com/paper/2204.06125/) (2022)
- [Adding Conditional Control to Text-to-Image Diffusion Models](https://mlsanity.com/paper/2302.05543/) (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)
- [Diffusion Model Alignment Using Direct Preference Optimization](https://mlsanity.com/paper/2311.12908/) (2023)
- [RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment](https://mlsanity.com/paper/2304.06767/) (2023)
- [DataComp: In search of the next generation of multimodal datasets](https://mlsanity.com/paper/2304.14108/) (2023)
- [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)
- [Flow Network based Generative Models for Non-Iterative Diverse Candidate Generation](https://mlsanity.com/paper/2106.04399/) (2021)
- [Unified Concept Editing in Diffusion Models](https://mlsanity.com/paper/2308.14761/) (2023)
- [Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discovery](https://mlsanity.com/paper/2302.03668/) (2023)
- [Ablating Concepts in Text-to-Image Diffusion Models](https://mlsanity.com/paper/2303.13516/) (2023)
- [MACE: Mass Concept Erasure in Diffusion Models](https://mlsanity.com/paper/2403.06135/) (2024)
- [The Stable Signature: Rooting Watermarks in Latent Diffusion Models](https://mlsanity.com/paper/2303.15435/) (2023)
- [GFlowNet Foundations](https://mlsanity.com/paper/2111.09266/) (2021)
- [Using Human Feedback to Fine-tune Diffusion Models without Any Reward Model](https://mlsanity.com/paper/2311.13231/) (2023)
- [Trajectory balance: Improved credit assignment in GFlowNets](https://mlsanity.com/paper/2201.13259/) (2022)
- [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)
- [MMA-Diffusion: MultiModal Attack on Diffusion Models](https://mlsanity.com/paper/2311.17516/) (2023)
- [Prompting4Debugging: Red-Teaming Text-to-Image Diffusion Models by Finding Problematic Prompts](https://mlsanity.com/paper/2309.06135/) (2023)
- [Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models](https://mlsanity.com/paper/2405.15234/) (2024)
- [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)
- [Receler: Reliable Concept Erasing of Text-to-Image Diffusion Models via Lightweight Erasers](https://mlsanity.com/paper/2311.17717/) (2023)

## Built on by (most cited first, 9 of 9)

- [Locality-Aware Continual Unlearning for Diffusion Models](https://mlsanity.com/paper/2512.02657/) (2025)
- [CGCE: Classifier-Guided Concept Erasure in Generative Models](https://mlsanity.com/paper/2511.05865/) (2025)
- [Orthogonal Concept Erasure for Diffusion Models](https://mlsanity.com/paper/2605.28902/) (2026)
- [GEM: Geometric Erasure by Contrastive Velocity Matching in Rectified Flows](https://mlsanity.com/paper/2606.00140/) (2026)
- [The Illusion of Forgetting: Attack Unlearned Diffusion via Initial Latent Variable Optimization](https://mlsanity.com/paper/2602.00175/) (2026)
- [Safe Few-Step Generation via Velocity Editing](https://mlsanity.com/paper/2606.23267/) (2026)
- [Obliviate: Erasing Concepts from Autoregressive Image Generation Models](https://mlsanity.com/paper/2606.28643/) (2026)
- [Your Unlearning Gives You Away: Identifying Erased Concepts in Diffusion Models](https://mlsanity.com/paper/2610.05601/) (2026)
- [TILDE: TILt-based Distributional Erasure for Concept Unlearning](https://mlsanity.com/paper/2607.06432/) (2026)

## Abstract (arXiv)

Erasing harmful or proprietary concepts from powerful text to image generators is an emerging safety requirement, yet current "concept erasure" techniques either collapse image quality, rely on brittle adversarial losses, or demand prohibitive retraining cycles. We trace these limitations to a myopic view of the denoising trajectories that govern diffusion based generation. We introduce EraseFlow, the first framework that casts concept unlearning as exploration in the space of denoising paths and optimizes it with GFlowNets equipped with the trajectory balance objective. By sampling entire trajectories rather than single end states, EraseFlow learns a stochastic policy that steers generation away from target concepts while preserving the model's prior. EraseFlow eliminates the need for carefully crafted reward models and by doing this, it generalizes effectively to unseen concepts and avoids hackable rewards while improving the performance. Extensive empirical results demonstrate that EraseFlow outperforms existing baselines and achieves an optimal trade off between performance and prior preservation.