Multimodal Large Language Models Red Teaming

Research areas:

• Volume 13, Issue 3, Nov. 2024

Year:

2024

Type of Publication:

Article

Keywords:

Red Teaming, Multimodal Large Language Models

Authors:

Dhruv Malik; Johnathan Yao; Boris Tkach

Journal:

IJAIM

Volume:

13

Number:

3

Pages:

1-6

Month:

November

ISSN:

2320-5121

Abstract:

Red teaming is a crucial process for identifying vulnerabilities and improving the robustness of multimodal large language models (LLMs). This survey reviews the existing literature on red teaming strategies for multimodal LLMs, evaluates their effectiveness, and provides recommendations for future research. We discuss the integration of various modalities, including text, image, and audio, and the unique challenges they present in red teaming.

Full text: IJAIM_681_FINAL.pdf