What happened
On 17 November 2026 CSA published an article detailing its STAR for AI certification program. It clarifies the two-level structure (Level 1 self-assessment using the AI Consensus Assessment Initiative Questionnaire / AI-CAIQ, and Level 2 independent third-party assessment via STAR for AI Certification or STAR for AI Attestation), the underlying AI Controls Matrix (AICM) which adds AI-specific controls (adversarial inputs, data poisoning, model integrity, explainability, human oversight, AI supply-chain transparency), the three in-scope AI service roles (Model Provider, Orchestrated Service Provider, Application Provider), the certification paths combining AICM with ISO/IEC 27001 and/or ISO/IEC 42001, and validity periods (Level 1 = 1 year; Level 2 Certification = 3 years; Level 2 Attestation = 1 year).
Why it matters
This is the authoritative practitioner-facing description of CSA's STAR for AI assurance scheme, which extends the world's largest cloud assurance program to AI. For AI service providers and their customers, it defines exactly what must be evidenced to obtain a STAR for AI listing/certification and how it maps onto ISO/IEC 27001/42001 and SOC 2, giving a concrete, auditable path to demonstrate AI security and governance beyond self-attestation.
Action needed
AI service providers should map their current controls against the AICM and plan a Level 1 self-assessment (AI-CAIQ) as a first step, then pursue Level 2 (certification or attestation) where customers/regulators require independent third-party assurance; AI consumers should add the STAR for AI registry to vendor assessment checklists.