Presentation + Paper
7 June 2024 An AI red team playbook
Anna Raney, Shiri Bendelac, Keith Manville, Mike Tan, Kureha Yamaguchi
Author Affiliations +
Abstract
As artificial intelligence (AI) continues to develop rapidly and influence numerous applications affecting billions of lives, it is crucial to form AI red teams whose objective is to identify AI-enabled system vulnerabilities before deployment to reduce likelihood or severity of real-world security risks. In response, we present a playbook to establish a formalized and repeatable process for AI red teaming. By describing the process as part of a larger framework known as Build-Attack-Defend (BAD), we define a collaborative process between the AI-enabled system development and security teams, as well as various stakeholders. Complementing An AI Blue Team Playbook, this paper contains the red teaming historical context, process, and lessons learned, serving as a starting point for proactively identifying weaknesses, enhancing the overall performance, security, and resilience of AI-enabled systems.
Conference Presentation
(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.
Anna Raney, Shiri Bendelac, Keith Manville, Mike Tan, and Kureha Yamaguchi "An AI red team playbook", Proc. SPIE 13054, Assurance and Security for AI-enabled Systems, 130540B (7 June 2024); https://doi.org/10.1117/12.3021906
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Artificial intelligence

Systems modeling

Machine learning

Defense and security

Back to Top