Teams of LLM Agents can Exploit Zero-Day Vulnerabilities
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_ | 1866917970368790528 |
|---|---|
| author | Zhu, Yuxuan Kellermann, Antony Gupta, Akul Li, Philip Fang, Richard Bindu, Rohan Kang, Daniel |
| author_facet | Zhu, Yuxuan Kellermann, Antony Gupta, Akul Li, Philip Fang, Richard Bindu, Rohan Kang, Daniel |
| contents | LLM agents have become increasingly sophisticated, especially in the realm of cybersecurity. Researchers have shown that LLM agents can exploit real-world vulnerabilities when given a description of the vulnerability and toy capture-the-flag problems. However, these agents still perform poorly on real-world vulnerabilities that are unknown to the agent ahead of time (zero-day vulnerabilities).
In this work, we show that teams of LLM agents can exploit real-world, zero-day vulnerabilities. Prior agents struggle with exploring many different vulnerabilities and long-range planning when used alone. To resolve this, we introduce HPTSA, a system of agents with a planning agent that can launch subagents. The planning agent explores the system and determines which subagents to call, resolving long-term planning issues when trying different vulnerabilities. We construct a benchmark of 14 real-world vulnerabilities and show that our team of agents improve over prior agent frameworks by up to 4.3X. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_01637 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Teams of LLM Agents can Exploit Zero-Day Vulnerabilities Zhu, Yuxuan Kellermann, Antony Gupta, Akul Li, Philip Fang, Richard Bindu, Rohan Kang, Daniel Multiagent Systems Artificial Intelligence I.2.7; D.4.6 LLM agents have become increasingly sophisticated, especially in the realm of cybersecurity. Researchers have shown that LLM agents can exploit real-world vulnerabilities when given a description of the vulnerability and toy capture-the-flag problems. However, these agents still perform poorly on real-world vulnerabilities that are unknown to the agent ahead of time (zero-day vulnerabilities). In this work, we show that teams of LLM agents can exploit real-world, zero-day vulnerabilities. Prior agents struggle with exploring many different vulnerabilities and long-range planning when used alone. To resolve this, we introduce HPTSA, a system of agents with a planning agent that can launch subagents. The planning agent explores the system and determines which subagents to call, resolving long-term planning issues when trying different vulnerabilities. We construct a benchmark of 14 real-world vulnerabilities and show that our team of agents improve over prior agent frameworks by up to 4.3X. |
| title | Teams of LLM Agents can Exploit Zero-Day Vulnerabilities |
| topic | Multiagent Systems Artificial Intelligence I.2.7; D.4.6 |
| url | https://arxiv.org/abs/2406.01637 |