Can Reinforcement Learning Unlock the Hidden Dangers in Aligned Large Language Models?
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909279949160448 |
|---|---|
| author | Karkevandi, Mohammad Bahrami Vishwamitra, Nishant Najafirad, Peyman |
| author_facet | Karkevandi, Mohammad Bahrami Vishwamitra, Nishant Najafirad, Peyman |
| contents | Large Language Models (LLMs) have demonstrated impressive capabilities in natural language tasks, but their safety and morality remain contentious due to their training on internet text corpora. To address these concerns, alignment techniques have been developed to improve the public usability and safety of LLMs. Yet, the potential for generating harmful content through these models seems to persist. This paper explores the concept of jailbreaking LLMs-reversing their alignment through adversarial triggers. Previous methods, such as soft embedding prompts, manually crafted prompts, and gradient-based automatic prompts, have had limited success on black-box models due to their requirements for model access and for producing a low variety of manually crafted prompts, making them susceptible to being blocked. This paper introduces a novel approach using reinforcement learning to optimize adversarial triggers, requiring only inference API access to the target model and a small surrogate model. Our method, which leverages a BERTScore-based reward function, enhances the transferability and effectiveness of adversarial triggers on new black-box models. We demonstrate that this approach improves the performance of adversarial triggers on a previously untested language model. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_02651 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Can Reinforcement Learning Unlock the Hidden Dangers in Aligned Large Language Models? Karkevandi, Mohammad Bahrami Vishwamitra, Nishant Najafirad, Peyman Computation and Language Artificial Intelligence Cryptography and Security Large Language Models (LLMs) have demonstrated impressive capabilities in natural language tasks, but their safety and morality remain contentious due to their training on internet text corpora. To address these concerns, alignment techniques have been developed to improve the public usability and safety of LLMs. Yet, the potential for generating harmful content through these models seems to persist. This paper explores the concept of jailbreaking LLMs-reversing their alignment through adversarial triggers. Previous methods, such as soft embedding prompts, manually crafted prompts, and gradient-based automatic prompts, have had limited success on black-box models due to their requirements for model access and for producing a low variety of manually crafted prompts, making them susceptible to being blocked. This paper introduces a novel approach using reinforcement learning to optimize adversarial triggers, requiring only inference API access to the target model and a small surrogate model. Our method, which leverages a BERTScore-based reward function, enhances the transferability and effectiveness of adversarial triggers on new black-box models. We demonstrate that this approach improves the performance of adversarial triggers on a previously untested language model. |
| title | Can Reinforcement Learning Unlock the Hidden Dangers in Aligned Large Language Models? |
| topic | Computation and Language Artificial Intelligence Cryptography and Security |
| url | https://arxiv.org/abs/2408.02651 |