The Risk Discovery Protocol for AI Assurance (RDP-AIA) is a standardized, multi-phase protocol that enables proactive discovery of use-case specific, mission-critical and sociotechnical risks in AI-enabled systems.
Risk Discovery Protocol for AI Assurance (v1.0)
Download Publication
In collaboration with Stakeholders, the protocol is designed to facilitate decomposition of a Stakeholder’s mission problem and AI-enabled solution, identification of existing assurance concerns with consequential mission-related or societal impacts, and to broaden Stakeholder’s aperture on assurance-related risk assessment in a way that facilitates discovery and prioritization of new and/or emergent AI assurance risks. The RDP-AIA leverages the AI Assurance (AIA) Landscape (Ward et al., 2024b), a synthesis of 50+ AI assurance frameworks that captures a comprehensive set of AIA needs and requirements—11 AIA categories, 66 specific assurance needs—captured in a broad range of existing frameworks, including NIST’s AI RMF. Collectively, the RDP-AIA and AIA Landscape support exploration of assurance needs to maximize use of AI-enhanced systems while minimizing the risks to the sociotechnical systems in which they are deployed.