Unveiling the Safety of GPT-4o: An Empirical Study using Jailbreak Attacks

TL;DR

This paper rigorously evaluates GPT-4o's safety against multi-modal jailbreak attacks across text, speech, and image, revealing improved safety in text but new vulnerabilities in audio, highlighting the need for better safety measures.

Contribution

First comprehensive empirical study of GPT-4o's safety against jailbreak attacks across multiple modalities, providing critical insights into its robustness and vulnerabilities.

Findings

GPT-4o shows enhanced safety in text modality jailbreaks

Audio modality introduces new attack vectors

Existing black-box attack methods are largely ineffective

Abstract

The recent release of GPT-4o has garnered widespread attention due to its powerful general capabilities. While its impressive performance is widely acknowledged, its safety aspects have not been sufficiently explored. Given the potential societal impact of risky content generated by advanced generative AI such as GPT-4o, it is crucial to rigorously evaluate its safety. In response to this question, this paper for the first time conducts a rigorous evaluation of GPT-4o against jailbreak attacks. Specifically, this paper adopts a series of multi-modal and uni-modal jailbreak attacks on 4 commonly used benchmarks encompassing three modalities (ie, text, speech, and image), which involves the optimization of over 4,000 initial text queries and the analysis and statistical evaluation of nearly 8,000+ response on GPT-4o. Our extensive experiments reveal several novel observations: (1) In contrast to the previous version (such as GPT-4V), GPT-4o has enhanced safety in the context of text modality jailbreak; (2) The newly introduced audio modality opens up new attack vectors for jailbreak attacks on GPT-4o; (3) Existing black-box multimodal jailbreak attack methods are largely ineffective against GPT-4o and GPT-4V. These findings provide critical insights into the safety implications of GPT-4o and underscore the need for robust alignment guardrails in large models. Our code is available at \url{https://github.com/NY1024/Jailbreak_GPT4o}.