Thames Kay Consultancy LLCTest your AI against 11 governance principles.
AI Verify, developed by Singapore's IMDA, is a testing framework that helps organisations assess AI systems against eleven internationally-recognised governance principles — consistent with the ASEAN, EU, OECD and US approaches. It pairs a structured process checklist with technical tests, giving you a defensible, evidence-based way to demonstrate responsible AI.
Outcome → Process → Evidence.
Outcome
Describe the outcome you want to achieve for each principle.
Process
The steps you take to achieve that outcome.
Evidence
The documentary, quantitative and qualitative proof that validates the process.
Internationally-recognised, audit-ready.
Transparency
Responsible disclosure so those affected can understand the outcome.
Explainability
Ability to explain how the AI reached its result.
Repeatability / Reproducibility
Consistent results under the same conditions.
Safety
The system behaves as intended and fails safely.
Security
Resilience against adversarial and malicious threats.
Robustness
Reliable performance across varied and unexpected inputs.
Fairness
No unintended bias against individuals or groups.
Data Governance
Sound provenance, quality and management of data.
Accountability
Clear ownership and responsibility for AI decisions.
Human Agency & Oversight
Meaningful human control over the system.
Inclusive Growth, Societal & Environmental Well-being
Broader benefit beyond the immediate use case.
Three audiences, one framework.
- AI system owners / developers — to demonstrate responsible AI governance in practice
- Internal compliance teams — to ensure responsible practices have actually been implemented
- External auditors — to validate a client's implementation of responsible AI
Independent summary of the AI Verify Testing Framework, developed by Singapore's Infocomm Media Development Authority (IMDA) and the AI Verify Foundation. Principles and structure are theirs; the commentary is ours.
Visit the AI Verify FoundationMake "responsible AI" something you can evidence.
We run AI Verify-aligned assessments and assemble the outcome-process-evidence record — so your responsible-AI claims hold up to an auditor.
Book an assessment