# Gaussian Core LoRA: Distribution-Aware Dynamic Adaptation for Broad Concept Erasure

arXiv: https://arxiv.org/abs/2609.01433 · HTML: https://mlsanity.com/paper/2609.01433/

## 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 7 of its 39 references that are in the map.
- First posted on arXiv 1 Sep 2026.

## Placed in

- Diffusion models / [Robustness, privacy and adversarial aspects](https://mlsanity.com/diffusion/robustness-privacy-and-adversarial-aspects/)

## Builds on (most cited first, 7 of 7)

- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [Red-Teaming the Stable Diffusion Safety Filter](https://mlsanity.com/paper/2210.04610/) (2022)
- [Prompting4Debugging: Red-Teaming Text-to-Image Diffusion Models by Finding Problematic Prompts](https://mlsanity.com/paper/2309.06135/) (2023)
- [UnlearnCanvas: Stylized Image Dataset for Enhanced Machine Unlearning Evaluation in Diffusion Models](https://mlsanity.com/paper/2402.11846/) (2024)
- [Separable Multi-Concept Erasure from Diffusion Models](https://mlsanity.com/paper/2402.05947/) (2024)
- [DyME: Dynamic Multi-Concept Erasure in Diffusion Models with Bi-Level Orthogonal LoRA Adaptation](https://mlsanity.com/paper/2509.21433/) (2025)

## Abstract (arXiv)

Concept erasure aims to suppress unsafe, privacy-sensitive, or undesirable generations in text-to-image diffusion models while preserving benign semantics, visual quality, and deployment efficiency. Existing adapter-based methods, such as Low-Rank Adaptation (LoRA), typically freeze the diffusion backbone and learn lightweight parameter updates to steer generation away from target semantics. However, these methods usually assign a static semantic erasure direction to each target concept. This assumption is overly coarse for broad and complex target concepts, since a concept often contains multiple latent semantic prototypes involving different objects, scenes, or relations, and requires different local erasure directions. A single LoRA update averages these heterogeneous erasure demands, leading to under-erasure on difficult prototypes and over-editing of nearby benign semantics. To address this limitation, we propose Gaussian Core LoRA, a distribution-aware low-rank adaptation framework. It fits a Gaussian mixture model in the prompt feature space to estimate latent semantic prototypes within the target concept. During inference, each input prompt is projected into this feature space to compute its Gaussian posterior responsibilities, which condition the core generator to produce a prompt-specific, norm-bounded residual reconfiguration of the shared LoRA rank space. This enables prototype-adaptive erasure with a single lightweight adapter. Compared with the strongest baseline on each metric, Gaussian Core LoRA reduces average Attack Success Rate (ASR) by 7.95%, lowers COCO Fr'echet Inception Distance (FID) by 14.72%, and improves CLIP Score by 4.98%. Further experiments show robustness to adversarial prompts, scalability to multi-identity and multi-style erasure, and compatibility with SDXL and FLUX.