CSA AICM v1.1 AI assurance
securityv1/01 Views
/02 About
Model AI use approval, lifecycle evidence, provenance, inference controls and continuous evaluation
Purpose: Model AI use approval, lifecycle evidence, provenance, inference controls and continuous evaluation. Reading the diagrams: Architecture landscape contains the complete semantic model. Governance and delivery isolates organizational decisions and enabling services. Delivery and assurance isolates runtime capabilities, controls, stewardship and evidence. Adoption: this is a reference architecture, not proof of implementation or regulatory compliance. Replace generic roles with actual owners and solutions. Record decisions, scope, constraints, interfaces, data classification, implementation status and evidence. Verify all applicable policies and control objectives against the current official source. Source and attribution: original Arq interpretation informed by https://cloudsecurityalliance.org/artifacts/ai-controls-matrix-v1-1. No official source artwork is reproduced, and no endorsement or certification by the source publisher is implied.
Published by Lattix · 12 elements · 13 relationships · validated on publish
/03 Contents
- Role
- AI risk owner
- Process
- AI use risk assessment, Model lifecycle governance, AI incident management
- Policy
- AI acceptable-use policy
- Data Store
- Training provenance, Model registry, AI assurance evidence
- Application Component
- Model evaluation, Inference policy gateway, Prompt and tool safeguards, Model telemetry