Stochastic Monkeys at Play: Random Augmentations Cheaply Break LLM Safety Alignment
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866909416628944896 |
|---|---|
| author | Vega, Jason Huang, Junsheng Zhang, Gaokai Kang, Hangoo Zhang, Minjia Singh, Gagandeep |
| author_facet | Vega, Jason Huang, Junsheng Zhang, Gaokai Kang, Hangoo Zhang, Minjia Singh, Gagandeep |
| contents | Safety alignment of Large Language Models (LLMs) has recently become a critical objective of model developers. In response, a growing body of work has been investigating how safety alignment can be bypassed through various jailbreaking methods, such as adversarial attacks. However, these jailbreak methods can be rather costly or involve a non-trivial amount of creativity and effort, introducing the assumption that malicious users are high-resource or sophisticated. In this paper, we study how simple random augmentations to the input prompt affect safety alignment effectiveness in state-of-the-art LLMs, such as Llama 3 and Qwen 2. We perform an in-depth evaluation of 17 different models and investigate the intersection of safety under random augmentations with multiple dimensions: augmentation type, model size, quantization, fine-tuning-based defenses, and decoding strategies (e.g., sampling temperature). We show that low-resource and unsophisticated attackers, i.e. $\textit{stochastic monkeys}$, can significantly improve their chances of bypassing alignment with just 25 random augmentations per prompt. Source code and data: https://github.com/uiuc-focal-lab/stochastic-monkeys/ |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_02785 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Stochastic Monkeys at Play: Random Augmentations Cheaply Break LLM Safety Alignment Vega, Jason Huang, Junsheng Zhang, Gaokai Kang, Hangoo Zhang, Minjia Singh, Gagandeep Machine Learning Artificial Intelligence Safety alignment of Large Language Models (LLMs) has recently become a critical objective of model developers. In response, a growing body of work has been investigating how safety alignment can be bypassed through various jailbreaking methods, such as adversarial attacks. However, these jailbreak methods can be rather costly or involve a non-trivial amount of creativity and effort, introducing the assumption that malicious users are high-resource or sophisticated. In this paper, we study how simple random augmentations to the input prompt affect safety alignment effectiveness in state-of-the-art LLMs, such as Llama 3 and Qwen 2. We perform an in-depth evaluation of 17 different models and investigate the intersection of safety under random augmentations with multiple dimensions: augmentation type, model size, quantization, fine-tuning-based defenses, and decoding strategies (e.g., sampling temperature). We show that low-resource and unsophisticated attackers, i.e. $\textit{stochastic monkeys}$, can significantly improve their chances of bypassing alignment with just 25 random augmentations per prompt. Source code and data: https://github.com/uiuc-focal-lab/stochastic-monkeys/ |
| title | Stochastic Monkeys at Play: Random Augmentations Cheaply Break LLM Safety Alignment |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2411.02785 |