Skip to content
GOVERNED AUTONOMY

AI governance readiness assessment

A free research diagnostic for private banks in Singapore.

Review your institution's reported practices against 193 assessment criteria across 7 domains, drawn from a defined corpus of 13 source instruments. The results support research and internal reflection; they do not establish regulatory compliance.

Short diagnostic Free · 35 priority criteria · approximately 25–30 minutes

Extended diagnostic Free · all 193 criteria · approximately 90–120 minutes

Both scopes use the same method and offer results in your browser and an optional PDF. No account or email address is required. Answers are saved on your device; requesting a PDF sends them to the report server. Read the privacy policy.

Seven assessment domains

Model Governance & Validation

115 CRITERIA

Full AI lifecycle from identification through decommissioning. Governance structure, capability, third-party AI, and generative AI controls.

Data Governance & Privacy

15 CRITERIA

Personal data in AI systems: consent, data protection, privacy-by-design, anonymisation, bias assessment, and third-party data processing.

Client-Facing AI & Suitability

14 CRITERIA

AI systems interacting with clients: algorithm governance, suitability of advice, fair dealing, customer transparency, and redress.

Explainability & Fairness

11 CRITERIA

Responsible AI principles: fairness definitions, protected attributes, bias detection, explainability methods, and agentic AI governance.

Outsourcing & Third-Party AI

12 CRITERIA

AI from external providers: governance framework, due diligence, outsourcing agreements, concentration risk, and lifecycle management.

Operational Resilience & Cybersecurity

12 CRITERIA

Technology and security infrastructure: IT governance, access control, audit logging, incident management, and AI-specific security.

Governance Structure & Accountability

14 CRITERIA

Institutional governance architecture: AI management system, risk appetite, AI risk culture, operating model, and skills and knowledge.

Source corpus

The diagnostic uses selected provisions from these 13 instruments. It includes proposals and voluntary frameworks to support research; those sources do not create current legal obligations. Instrument names link to the publishers’ source pages or directories.

  1. P017 – Proposed Guidelines on AI Risk Management

    MAS · November 2025 proposal; consultation closed 31 January 2026

    CONSULTATION

    58 criteria

  2. METHODOLOGY

    32 criteria

  3. ISO/IEC 42001:2023

    ISO/IEC · Edition 1, December 2023

    Publisher overview; the full standard requires licensed access.

    ASSURANCE

    19 criteria

  4. FEAT Principles (2018)

    MAS · 2018 corpus edition

    METHODOLOGY

    17 criteria

  5. OBSERVED PRACTICE

    16 criteria

  6. TRM Guidelines 2021

    MAS · January 2021

    SUPERVISORY

    13 criteria

  7. METHODOLOGY

    11 criteria

  8. SUPERVISORY

    8 criteria

  9. SUPERVISORY

    7 criteria

  10. Veritas Assessment Methodology

    MAS/Industry · Edition not specified in the register

    The link opens the initiative page; edition-level mapping remains to be confirmed.

    METHODOLOGY

    5 criteria

  11. Fair Dealing Guidelines (May 2024)

    MAS · May 2024

    The link opens the MAS guidelines directory.

    SUPERVISORY

    3 criteria

  12. IMDA Model AI Governance Framework for Agentic AI

    IMDA · January 2026 corpus edition (v1.0)

    January launch page. The publisher has since issued a revised framework.

    METHODOLOGY

    2 criteria

  13. P004 – Proposed Guidelines on Third-Party Risk Management

    MAS · March 2026 proposal; consultation closed 20 April 2026

    CONSULTATION

    2 criteria

About this diagnostic

Adrien Pesa developed this diagnostic within Governed Autonomy, his independent, self-funded and non-commercial research programme. The programme has no entity, clients or revenue. The diagnostic is not affiliated with or endorsed by any regulator.

The source framework distinguishes five normative tiers and a separate observed-practice class. The current corpus contains no statutory criteria. Proposed guidance, voluntary methodologies and assurance standards are distinguished from supervisory expectations.

Read the methodology · Explore the research programme

Research feedback and corrections are welcome at contact@governed-autonomy.com.