AVISE: Framework for Evaluating the Security of AI Systems

TL;DR

AVISE is an open-source framework designed to systematically evaluate and identify security vulnerabilities in AI systems, demonstrated through a new attack method and a security evaluation test for language models.

Contribution

Introduces AVISE, a modular framework for AI security assessment, including a novel attack extension and an automated jailbreak vulnerability testing tool.

Findings

All evaluated language models are vulnerable to the augmented Red Queen attack.

The Security Evaluation Test (SET) achieved 92% accuracy and 0.91 F1-score.

AVISE enables reproducible and extensible AI security evaluations.

Abstract

As artificial intelligence (AI) systems are increasingly deployed across critical domains, their security vulnerabilities pose growing risks of high-profile exploits and consequential system failures. Yet systematic approaches to evaluating AI security remain underdeveloped. In this paper, we introduce AVISE (AI Vulnerability Identification and Security Evaluation), a modular open-source framework for identifying vulnerabilities in and evaluating the security of AI systems and models. As a demonstration of the framework, we extend the theory-of-mind-based multi-turn Red Queen attack into an Adversarial Language Model (ALM) augmented attack and develop an automated Security Evaluation Test (SET) for discovering jailbreak vulnerabilities in language models. The SET comprises 25 test cases and an Evaluation Language Model (ELM) that determines whether each test case was able to jailbreak the target model, achieving 92% accuracy, an F1-score of 0.91, and a Matthews correlation coefficient of 0.83. We evaluate nine recently released language models of diverse sizes with the SET and find that all are vulnerable to the augmented Red Queen attack to varying degrees. AVISE provides researchers and industry practitioners with an extensible foundation for developing and deploying automated SETs, offering a concrete step toward more rigorous and reproducible AI security evaluation.