Establishing Minimum Elements for Effective Vulnerability Management in AI Software

TL;DR

The paper addresses the need for robust vulnerability management in AI systems, highlighting that traditional vulnerability databases fail to capture AI-specific threats tied to data, models, and adaptive behavior. It proposes the Artificial Intelligence Vulnerability Database (AIVD) and associated framework, including AI-CWE, a minimum-element schema, and the AI Bill of Materials (AIBOM), to standardize disclosure, analysis, and mitigation. The work discusses AI-specific gaps in severity scoring, weakness enumeration, and mitigation, and presents a path for future challenges and improvements. Through a motivating example, it demonstrates practical use of AIVD for documenting and addressing vulnerabilities such as membership inference in cloud AI services.

Abstract

In the rapidly evolving field of artificial intelligence (AI), the identification, documentation, and mitigation of vulnerabilities are paramount to ensuring robust and secure systems. This paper discusses the minimum elements for AI vulnerability management and the establishment of an Artificial Intelligence Vulnerability Database (AIVD). It presents standardized formats and protocols for disclosing, analyzing, cataloging, and documenting AI vulnerabilities. It discusses how such an AI incident database must extend beyond the traditional scope of vulnerabilities by focusing on the unique aspects of AI systems. Additionally, this paper highlights challenges and gaps in AI Vulnerability Management, including the need for new severity scores, weakness enumeration systems, and comprehensive mitigation strategies specifically designed to address the multifaceted nature of AI vulnerabilities.

Figures (1)

• Figure 1: An overview of the proposed AIBOM