NeuRel-Attack: Neuron Relearning for Safety Disalignment in Large Language Models
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910921933193216 |
|---|---|
| author | Zhou, Yi Xing, Wenpeng Kong, Dezhang Lin, Changting Han, Meng |
| author_facet | Zhou, Yi Xing, Wenpeng Kong, Dezhang Lin, Changting Han, Meng |
| contents | Safety alignment in large language models (LLMs) is achieved through fine-tuning mechanisms that regulate neuron activations to suppress harmful content. In this work, we propose a novel approach to induce disalignment by identifying and modifying the neurons responsible for safety constraints. Our method consists of three key steps: Neuron Activation Analysis, where we examine activation patterns in response to harmful and harmless prompts to detect neurons that are critical for distinguishing between harmful and harmless inputs; Similarity-Based Neuron Identification, which systematically locates the neurons responsible for safe alignment; and Neuron Relearning for Safety Removal, where we fine-tune these selected neurons to restore the model's ability to generate previously restricted responses. Experimental results demonstrate that our method effectively removes safety constraints with minimal fine-tuning, highlighting a critical vulnerability in current alignment techniques. Our findings underscore the need for robust defenses against adversarial fine-tuning attacks on LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_21053 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | NeuRel-Attack: Neuron Relearning for Safety Disalignment in Large Language Models Zhou, Yi Xing, Wenpeng Kong, Dezhang Lin, Changting Han, Meng Machine Learning Artificial Intelligence Safety alignment in large language models (LLMs) is achieved through fine-tuning mechanisms that regulate neuron activations to suppress harmful content. In this work, we propose a novel approach to induce disalignment by identifying and modifying the neurons responsible for safety constraints. Our method consists of three key steps: Neuron Activation Analysis, where we examine activation patterns in response to harmful and harmless prompts to detect neurons that are critical for distinguishing between harmful and harmless inputs; Similarity-Based Neuron Identification, which systematically locates the neurons responsible for safe alignment; and Neuron Relearning for Safety Removal, where we fine-tune these selected neurons to restore the model's ability to generate previously restricted responses. Experimental results demonstrate that our method effectively removes safety constraints with minimal fine-tuning, highlighting a critical vulnerability in current alignment techniques. Our findings underscore the need for robust defenses against adversarial fine-tuning attacks on LLMs. |
| title | NeuRel-Attack: Neuron Relearning for Safety Disalignment in Large Language Models |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2504.21053 |