# Why Instruction-Based Unlearning Fails in Diffusion Models?

arXiv: https://arxiv.org/abs/2604.01514 · HTML: https://mlsanity.com/paper/2604.01514/

## 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 10 of its 15 references that are in the map.
- First posted on arXiv 2 Apr 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, 10 of 10)

- [Diffusion Models in Vision: A Survey](https://mlsanity.com/paper/2209.04747/) (2022)
- [MM-Diffusion: Learning Multi-Modal Diffusion Models for Joint Audio and Video Generation](https://mlsanity.com/paper/2212.09478/) (2022)
- [In-Context Unlearning: Language Models as Few Shot Unlearners](https://mlsanity.com/paper/2310.07579/) (2023)
- [DDFM: Denoising Diffusion Model for Multi-Modality Image Fusion](https://mlsanity.com/paper/2303.06840/) (2023)
- [Unlearning Concepts in Diffusion Model via Concept Domain Correction and Concept Preserving Gradient](https://mlsanity.com/paper/2405.15304/) (2024)
- [Meta-Unlearning on Diffusion Models: Preventing Relearning Unlearned Concepts](https://mlsanity.com/paper/2410.12777/) (2024)
- [Erasing Concepts from Text-to-Image Diffusion Models with Few-shot Unlearning](https://mlsanity.com/paper/2405.07288/) (2024)
- [Can CLIP Count Stars? An Empirical Study on Quantity Bias in CLIP](https://mlsanity.com/paper/2409.15035/) (2024)
- [Model Integrity when Unlearning with T2I Diffusion Models](https://mlsanity.com/paper/2411.02068/) (2024)
- [Targeted Forgetting of Image Subgroups in CLIP Models](https://mlsanity.com/paper/2506.03117/) (2025)

## Abstract (arXiv)

Instruction-based unlearning has proven effective for modifying the behavior of large language models at inference time, but whether this paradigm extends to other generative models remains unclear. In this work, we investigate instruction-based unlearning in diffusion-based image generation models and show, through controlled experiments across multiple concepts and prompt variants, that diffusion models systematically fail to suppress targeted concepts when guided solely by natural-language unlearning instructions. By analyzing both the CLIP text encoder and cross-attention dynamics during the denoising process, we find that unlearning instructions do not induce sustained reductions in attention to the targeted concept tokens, causing the targeted concept representations to persist throughout generation. These results reveal a fundamental limitation of prompt-level instruction in diffusion models and suggest that effective unlearning requires interventions beyond inference-time language control.