Handbook
The GovernanceOS AI Governance Handbook
A free, practical AI governance handbook for regulated institutions and the practitioners who advise them: the operating model, the lifecycle, the evidence, regulation by region, and governing more capable AI.
Contents
Thirteen chapters, from the first principle to superintelligence.
Chapters are published one at a time. Until a chapter has its own page, its entry opens the public page that covers the topic today. Seven entries open the Defensible AI Library, where every document is described for anyone and opens after sign-up.
- 01What AI governance isAI ethics, AI regulation and AI governance, and the five core pillars.Covered today on About
- 02The operating modelThe charter, who decides what, the meeting records and the escalation paths.Covered today in the Defensible AI Library; its documents open after sign-up
- 03Knowing your AIThe inventory: every model, agent and vendor service, with an owner and a review date.Covered today on Register tier
- 04Assessing a systemWho a system affects, what it is built from, where its data comes from, and its model risk.Covered today in the Defensible AI Library; its documents open after sign-up
- 05Passing a gateFive gates from data and design to operation, each passed by one named approver on its evidence.Covered today in the Defensible AI Library; its documents open after sign-up
- 06Third-party AIAssessing a vendor’s AI on the same terms as your own, and keeping approval through its releases.Covered today in the Defensible AI Library; its documents open after sign-up
- 07When AI failsFrom the first signal to the review: classify, contain, reconstruct and close.Covered today in the Defensible AI Library; its documents open after sign-up
- 08Evidence an examiner can checkRecords that say who decided, on what and when, and how a pack is checked without the vendor.Covered today on Verify a pack
- 09Regulation by regionOSFI Guideline E-23 and the AMF guideline, US federal and state rules, six GCC states, the EU AI Act stage by stage, the UK, Switzerland, Asia-Pacific and the international standards.Covered today on Regulation rows
- 10Sharia AI governanceTechnical validation and Sharia review as two tracks over the same records, closed together.Covered today in the Defensible AI Library; its documents open after sign-up
- 11Proving itThe maturity position, the board’s quarterly briefing, the regulatory mapping and the audit trail.Covered today in the Defensible AI Library; its documents open after sign-up
- 12Governing more capable AIControls designed to hold as capability grows, up to superintelligence, and where each would fail.Covered today on About
- 13The peopleThe roles that run AI governance, and the certification path for people and institutions.Covered today on AI governance certification
By role
Where to start.
- CRO and executives
- Chapters 1, 2, 11 and 12
- Compliance leads
- Chapters 3, 5, 9 and 11
- Validators and model risk
- Chapters 4, 5 and 8
- Procurement and vendor owners
- Chapters 6 and 9
- Sharia board liaisons
- Chapters 2, 9 and 10
- Practitioners and consultants
- Every chapter, starting with 1, 3 and 5
Questions
Questions about the handbook
What is an AI governance handbook?
A practical guide to running AI governance: who owns each system, who may approve it, on what evidence, and what happens when it fails. This one is written for regulated institutions and the practitioners who advise them.
Is the GovernanceOS AI Governance Handbook free?
Yes, to read. Every entry opens a public page. Where an entry opens the Defensible AI Library, each document is described for anyone, and the documents themselves open after sign-up.
Who is it for?
CROs, compliance leads, validators, procurement and vendor owners, Sharia board liaisons, and practitioners who run governance for several institutions.
How is it different from a book on AI governance?
It is built for practice on records. Each entry links the documents, regulation rows and certification levels that put the topic to work, and the regulation rows carry the date each was checked.
Which regulations does it cover?
OSFI Guideline E-23 and Quebec’s AMF guideline, US federal and state rules, and six GCC states, with the common standards: the NIST AI RMF, ISO/IEC 42001 and the OECD AI Principles.
How are chapters published?
One at a time. Until a chapter has its own page, its entry links to where GovernanceOS already covers the topic.