# Erasing CLIP Memories: Non-Destructive, Data-Free Zero-Shot class Unlearning in CLIP Models

arXiv: https://arxiv.org/abs/2512.14137 · HTML: https://mlsanity.com/paper/2512.14137/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Machine unlearning, one of 12 directions in LLM safety and security.
- Builds on 9 of its 17 references that are in the map.
- First posted on arXiv 16 Dec 2025.

## Placed in

- LLM safety and security / [Machine unlearning](https://mlsanity.com/safety/machine-unlearning/)

## Builds on (most cited first, 9 of 9)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Machine Unlearning](https://mlsanity.com/paper/1912.03817/) (2019)
- [Null It Out: Guarding Protected Attributes by Iterative Nullspace Projection](https://mlsanity.com/paper/2004.07667/) (2020)
- [SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation](https://mlsanity.com/paper/2310.12508/) (2023)
- [Machine Unlearning for Image-to-Image Generative Models](https://mlsanity.com/paper/2402.00351/) (2024)
- [Online Zero-Shot Classification with CLIP](https://mlsanity.com/paper/2408.13320/) (2024)
- [CLIPErase: Efficient Unlearning of Visual-Textual Associations in CLIP](https://mlsanity.com/paper/2410.23330/) (2024)
- [Zero-Shot Class Unlearning in CLIP with Synthetic Samples](https://mlsanity.com/paper/2407.07485/) (2024)
- [On the Limitations and Prospects of Machine Unlearning for Generative AI](https://mlsanity.com/paper/2408.00376/) (2024)

## Built on by (most cited first, 1 of 1)

- [Toward Fine-Grained Forgetting:Attribute Unlearning for Multimodal Large Language Models](https://mlsanity.com/paper/2608.01008/) (2026)

## Abstract (arXiv)

We introduce a novel, closed-form approach for selective unlearning in multimodal models, specifically targeting pretrained models such as CLIP. Our method leverages nullspace projection to erase the target class information embedded in the final projection layer, without requiring any retraining or the use of images from the forget set. By computing an orthonormal basis for the subspace spanned by target text embeddings and projecting these directions, we dramatically reduce the alignment between image features and undesired classes. Unlike traditional unlearning techniques that rely on iterative fine-tuning and extensive data curation, our approach is both computationally efficient and surgically precise. This leads to a pronounced drop in zero-shot performance for the target classes while preserving the overall multimodal knowledge of the model. Our experiments demonstrate that even a partial projection can balance between complete unlearning and retaining useful information, addressing key challenges in model decontamination and privacy preservation.