Guardado en:
Detalles Bibliográficos
Autores principales: Pavlova, Maya, Brinkman, Erik, Iyer, Krithika, Albiero, Vitor, Bitton, Joanna, Nguyen, Hailey, Li, Joe, Ferrer, Cristian Canton, Evtimov, Ivan, Grattafiori, Aaron
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2410.01606
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913527218831360
author Pavlova, Maya
Brinkman, Erik
Iyer, Krithika
Albiero, Vitor
Bitton, Joanna
Nguyen, Hailey
Li, Joe
Ferrer, Cristian Canton
Evtimov, Ivan
Grattafiori, Aaron
author_facet Pavlova, Maya
Brinkman, Erik
Iyer, Krithika
Albiero, Vitor
Bitton, Joanna
Nguyen, Hailey
Li, Joe
Ferrer, Cristian Canton
Evtimov, Ivan
Grattafiori, Aaron
contents Red teaming assesses how large language models (LLMs) can produce content that violates norms, policies, and rules set during their safety training. However, most existing automated methods in the literature are not representative of the way humans tend to interact with AI models. Common users of AI models may not have advanced knowledge of adversarial machine learning methods or access to model internals, and they do not spend a lot of time crafting a single highly effective adversarial prompt. Instead, they are likely to make use of techniques commonly shared online and exploit the multiturn conversational nature of LLMs. While manual testing addresses this gap, it is an inefficient and often expensive process. To address these limitations, we introduce the Generative Offensive Agent Tester (GOAT), an automated agentic red teaming system that simulates plain language adversarial conversations while leveraging multiple adversarial prompting techniques to identify vulnerabilities in LLMs. We instantiate GOAT with 7 red teaming attacks by prompting a general-purpose model in a way that encourages reasoning through the choices of methods available, the current target model's response, and the next steps. Our approach is designed to be extensible and efficient, allowing human testers to focus on exploring new areas of risk while automation covers the scaled adversarial stress-testing of known risk territory. We present the design and evaluation of GOAT, demonstrating its effectiveness in identifying vulnerabilities in state-of-the-art LLMs, with an ASR@10 of 97% against Llama 3.1 and 88% against GPT-4 on the JailbreakBench dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2410_01606
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Automated Red Teaming with GOAT: the Generative Offensive Agent Tester
Pavlova, Maya
Brinkman, Erik
Iyer, Krithika
Albiero, Vitor
Bitton, Joanna
Nguyen, Hailey
Li, Joe
Ferrer, Cristian Canton
Evtimov, Ivan
Grattafiori, Aaron
Machine Learning
Artificial Intelligence
Red teaming assesses how large language models (LLMs) can produce content that violates norms, policies, and rules set during their safety training. However, most existing automated methods in the literature are not representative of the way humans tend to interact with AI models. Common users of AI models may not have advanced knowledge of adversarial machine learning methods or access to model internals, and they do not spend a lot of time crafting a single highly effective adversarial prompt. Instead, they are likely to make use of techniques commonly shared online and exploit the multiturn conversational nature of LLMs. While manual testing addresses this gap, it is an inefficient and often expensive process. To address these limitations, we introduce the Generative Offensive Agent Tester (GOAT), an automated agentic red teaming system that simulates plain language adversarial conversations while leveraging multiple adversarial prompting techniques to identify vulnerabilities in LLMs. We instantiate GOAT with 7 red teaming attacks by prompting a general-purpose model in a way that encourages reasoning through the choices of methods available, the current target model's response, and the next steps. Our approach is designed to be extensible and efficient, allowing human testers to focus on exploring new areas of risk while automation covers the scaled adversarial stress-testing of known risk territory. We present the design and evaluation of GOAT, demonstrating its effectiveness in identifying vulnerabilities in state-of-the-art LLMs, with an ASR@10 of 97% against Llama 3.1 and 88% against GPT-4 on the JailbreakBench dataset.
title Automated Red Teaming with GOAT: the Generative Offensive Agent Tester
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2410.01606