Cybersecurity AI Benchmark (CAIBench): A Meta-Benchmark for Evaluating Cybersecurity AI Agents

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sanz-Gómez, María, Mayoral-Vilches, Víctor, Balassone, Francesco, Navarrete-Lozano, Luis Javier, Chavez, Cristóbal R. J. Veas, de Torres, Maite del Mundo
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912673977860096
author Sanz-Gómez, María
Mayoral-Vilches, Víctor
Balassone, Francesco
Navarrete-Lozano, Luis Javier
Chavez, Cristóbal R. J. Veas
de Torres, Maite del Mundo
author_facet Sanz-Gómez, María
Mayoral-Vilches, Víctor
Balassone, Francesco
Navarrete-Lozano, Luis Javier
Chavez, Cristóbal R. J. Veas
de Torres, Maite del Mundo
contents Cybersecurity spans multiple interconnected domains, complicating the development of meaningful, labor-relevant benchmarks. Existing benchmarks assess isolated skills rather than integrated performance. We find that pre-trained knowledge of cybersecurity in LLMs does not imply attack and defense abilities, revealing a gap between knowledge and capability. To address this limitation, we present the Cybersecurity AI Benchmark (CAIBench), a modular meta-benchmark framework that allows evaluating LLM models and agents across offensive and defensive cybersecurity domains, taking a step towards meaningfully measuring their labor-relevance. CAIBench integrates five evaluation categories, covering over 10,000 instances: Jeopardy-style CTFs, Attack and Defense CTFs, Cyber Range exercises, knowledge benchmarks, and privacy assessments. Key novel contributions include systematic simultaneous offensive-defensive evaluation, robotics-focused cybersecurity challenges (RCTF2), and privacy-preserving performance assessment (CyberPII-Bench). Evaluation of state-of-the-art AI models reveals saturation on security knowledge metrics (~70\% success) but substantial degradation in multi-step adversarial (A\&D) scenarios (20-40\% success), or worse in robotic targets (22\% success). The combination of framework scaffolding and LLM model choice significantly impacts performance; we find that proper matches improve up to 2.6$\times$ variance in Attack and Defense CTFs. These results demonstrate a pronounced gap between conceptual knowledge and adaptive capability, emphasizing the need for a meta-benchmark.
format Preprint
id arxiv_https___arxiv_org_abs_2510_24317
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Cybersecurity AI Benchmark (CAIBench): A Meta-Benchmark for Evaluating Cybersecurity AI Agents
Sanz-Gómez, María
Mayoral-Vilches, Víctor
Balassone, Francesco
Navarrete-Lozano, Luis Javier
Chavez, Cristóbal R. J. Veas
de Torres, Maite del Mundo
Cryptography and Security
Cybersecurity spans multiple interconnected domains, complicating the development of meaningful, labor-relevant benchmarks. Existing benchmarks assess isolated skills rather than integrated performance. We find that pre-trained knowledge of cybersecurity in LLMs does not imply attack and defense abilities, revealing a gap between knowledge and capability. To address this limitation, we present the Cybersecurity AI Benchmark (CAIBench), a modular meta-benchmark framework that allows evaluating LLM models and agents across offensive and defensive cybersecurity domains, taking a step towards meaningfully measuring their labor-relevance. CAIBench integrates five evaluation categories, covering over 10,000 instances: Jeopardy-style CTFs, Attack and Defense CTFs, Cyber Range exercises, knowledge benchmarks, and privacy assessments. Key novel contributions include systematic simultaneous offensive-defensive evaluation, robotics-focused cybersecurity challenges (RCTF2), and privacy-preserving performance assessment (CyberPII-Bench). Evaluation of state-of-the-art AI models reveals saturation on security knowledge metrics (~70\% success) but substantial degradation in multi-step adversarial (A\&D) scenarios (20-40\% success), or worse in robotic targets (22\% success). The combination of framework scaffolding and LLM model choice significantly impacts performance; we find that proper matches improve up to 2.6$\times$ variance in Attack and Defense CTFs. These results demonstrate a pronounced gap between conceptual knowledge and adaptive capability, emphasizing the need for a meta-benchmark.
title Cybersecurity AI Benchmark (CAIBench): A Meta-Benchmark for Evaluating Cybersecurity AI Agents
topic Cryptography and Security
url https://arxiv.org/abs/2510.24317