LLM Agents can Autonomously Exploit One-day Vulnerabilities

TL;DR

The paper investigates whether autonomous LLM agents can exploit real-world, one-day vulnerabilities. Using a benchmark of 15 CVEs, a GPT-4-based agent with CVE descriptions achieves an 87% success rate, vastly outperforming GPT-3.5, open-source models, and vulnerability scanners that reach 0%. Removing the CVE descriptions drops GPT-4 performance to 7%, highlighting the critical role of vulnerability context. The study provides a cost analysis, discusses agent capabilities and limitations, and emphasizes the need for defensive considerations in deploying LLM agents in cybersecurity contexts. Overall, it demonstrates an emergent, high-risk capability in GPT-4 and urges careful governance and safety measures in practical deployments.

Abstract

LLMs have becoming increasingly powerful, both in their benign and malicious uses. With the increase in capabilities, researchers have been increasingly interested in their ability to exploit cybersecurity vulnerabilities. In particular, recent work has conducted preliminary studies on the ability of LLM agents to autonomously hack websites. However, these studies are limited to simple vulnerabilities. In this work, we show that LLM agents can autonomously exploit one-day vulnerabilities in real-world systems. To show this, we collected a dataset of 15 one-day vulnerabilities that include ones categorized as critical severity in the CVE description. When given the CVE description, GPT-4 is capable of exploiting 87% of these vulnerabilities compared to 0% for every other model we test (GPT-3.5, open-source LLMs) and open-source vulnerability scanners (ZAP and Metasploit). Fortunately, our GPT-4 agent requires the CVE description for high performance: without the description, GPT-4 can exploit only 7% of the vulnerabilities. Our findings raise questions around the widespread deployment of highly capable LLM agents.

Figures (1)

• Figure 1: System diagram of our LLM agent.