SPIN: Self-Supervised Prompt INjection
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916442702610432 |
|---|---|
| author | Zhou, Leon Yang, Junfeng Mao, Chengzhi |
| author_facet | Zhou, Leon Yang, Junfeng Mao, Chengzhi |
| contents | Large Language Models (LLMs) are increasingly used in a variety of important applications, yet their safety and reliability remain as major concerns. Various adversarial and jailbreak attacks have been proposed to bypass the safety alignment and cause the model to produce harmful responses. We introduce Self-supervised Prompt INjection (SPIN) which can detect and reverse these various attacks on LLMs. As our self-supervised prompt defense is done at inference-time, it is also compatible with existing alignment and adds an additional layer of safety for defense. Our benchmarks demonstrate that our system can reduce the attack success rate by up to 87.9%, while maintaining the performance on benign user requests. In addition, we discuss the situation of an adaptive attacker and show that our method is still resilient against attackers who are aware of our defense. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_13236 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | SPIN: Self-Supervised Prompt INjection Zhou, Leon Yang, Junfeng Mao, Chengzhi Computation and Language Artificial Intelligence Large Language Models (LLMs) are increasingly used in a variety of important applications, yet their safety and reliability remain as major concerns. Various adversarial and jailbreak attacks have been proposed to bypass the safety alignment and cause the model to produce harmful responses. We introduce Self-supervised Prompt INjection (SPIN) which can detect and reverse these various attacks on LLMs. As our self-supervised prompt defense is done at inference-time, it is also compatible with existing alignment and adds an additional layer of safety for defense. Our benchmarks demonstrate that our system can reduce the attack success rate by up to 87.9%, while maintaining the performance on benign user requests. In addition, we discuss the situation of an adaptive attacker and show that our method is still resilient against attackers who are aware of our defense. |
| title | SPIN: Self-Supervised Prompt INjection |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2410.13236 |