Dynamic Risk Assessments for Offensive Cybersecurity Agents
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918179564945408 |
|---|---|
| author | Wei, Boyi Stroebl, Benedikt Xu, Jiacen Zhang, Joie Li, Zhou Henderson, Peter |
| author_facet | Wei, Boyi Stroebl, Benedikt Xu, Jiacen Zhang, Joie Li, Zhou Henderson, Peter |
| contents | Foundation models are increasingly becoming better autonomous programmers, raising the prospect that they could also automate dangerous offensive cyber-operations. Current frontier model audits probe the cybersecurity risks of such agents, but most fail to account for the degrees of freedom available to adversaries in the real world. In particular, with strong verifiers and financial incentives, agents for offensive cybersecurity are amenable to iterative improvement by would-be adversaries. We argue that assessments should take into account an expanded threat model in the context of cybersecurity, emphasizing the varying degrees of freedom that an adversary may possess in stateful and non-stateful environments within a fixed compute budget. We show that even with a relatively small compute budget (8 H100 GPU Hours in our study), adversaries can improve an agent's cybersecurity capability on InterCode CTF by more than 40\% relative to the baseline -- without any external assistance. These results highlight the need to evaluate agents' cybersecurity risk in a dynamic manner, painting a more representative picture of risk. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_18384 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Dynamic Risk Assessments for Offensive Cybersecurity Agents Wei, Boyi Stroebl, Benedikt Xu, Jiacen Zhang, Joie Li, Zhou Henderson, Peter Cryptography and Security Artificial Intelligence Foundation models are increasingly becoming better autonomous programmers, raising the prospect that they could also automate dangerous offensive cyber-operations. Current frontier model audits probe the cybersecurity risks of such agents, but most fail to account for the degrees of freedom available to adversaries in the real world. In particular, with strong verifiers and financial incentives, agents for offensive cybersecurity are amenable to iterative improvement by would-be adversaries. We argue that assessments should take into account an expanded threat model in the context of cybersecurity, emphasizing the varying degrees of freedom that an adversary may possess in stateful and non-stateful environments within a fixed compute budget. We show that even with a relatively small compute budget (8 H100 GPU Hours in our study), adversaries can improve an agent's cybersecurity capability on InterCode CTF by more than 40\% relative to the baseline -- without any external assistance. These results highlight the need to evaluate agents' cybersecurity risk in a dynamic manner, painting a more representative picture of risk. |
| title | Dynamic Risk Assessments for Offensive Cybersecurity Agents |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2505.18384 |