Implementing AI RMF with Policy-as-Code Automation - Robert Ficcaglia, Anca Sailer & Vikas Agarwal

Implementing AI RMF with Policy-as-Code Automation - Robert Ficcaglia, SunStone Secure; Anca Sailer & Vikas Agarwal, IBM

This session will focus on AI Risk Assessment, Compliance Assurance, and Red Teaming for AI models and AI pipelines deployed on Kubernetes cloud native platforms. We will map the LinuxFoundation Trusted AI Principles of Reproducibility, Robustness, Equitability, Privacy, Explainability, Accountability, Transparency, and Security to the NIST AI RMF, and define a reusable framework for designing controls to implement these principles and requirements. We will show policy-as-code templates that enforce controls throughout the AI life cycle, and discuss how to report risks and show examples of compliance artifacts for Privacy and Bias validation. The session will be led by experienced AI and compliance practitioners who are implementing red teaming and AI safety assurance using Kubernetes and CNCF open source tools. This session will work through specific examples, and AI SMEs will provide feedback and suggestions regarding attendees’ questions and scenarios.