GovCompass
NL

From having AI controls to being demonstrably in control

Most organizations have AI policies. Far fewer can show their AI is governed, controlled, and compliant in practice. GovCompass is a free knowledge base that covers that whole distance: what AI governance is, how you stay in control, and what the EU AI Act requires.

Agentic AIacross all seven pillarsFairnessSafetyreliabilityPrivacySecurityrobustnessTransparencyexplainabilityAccountabilityHumanoversightResponsible AI

Explore the principle-to-evidence chain

Understand and design

AI governance

What AI governance is, and how to design it: the seven pillars of responsible AI, the governance layers, and what changes when AI becomes agentic.

View section
Operate and prove

AI in control

Having controls is not the same as being in control. Risk management, certification, and the evidence that shows your AI controls operate.

View section
The law and the layers

AI regulations & frameworks

The EU AI Act, article by article, kept current against the official text. Plus the frameworks around it: ISO/IEC 42001 and the NIST AI RMF.

View section
Learn and certify

Academy

Structured preparation for the IAPP AIGP exam: around 50 lessons across the four domains, practice questions, and exam guidance. Built on the same verified sources as everything on this site.

View section
What AI governance is

Governing AI is now a board-level responsibility

AI governance is how an organization keeps its AI lawful, fair and accountable, the policies, roles and controls that decide which systems are allowed, who is responsible, and how decisions are checked across the whole AI lifecycle.

It is no longer optional. The EU AI Act is in force, with phased deadlines through 2028, obligations for both providers and deployers, mandatory human oversight and documentation, and fines up to €35 million or 7% of global turnover. GovCompass is where you understand these obligations, free, and where you build the depth to apply them.

Start with AI governance How GovCompass works

  • In force now

    The EU AI Act applies, phased deadlines run through 2028.

  • Everyone in scope

    Both providers and deployers of AI carry obligations.

  • Real consequences

    Fines up to €35M or 7% of global annual turnover.

From the knowledge base

Recent analyses & guides

Free, plain-language answers on the EU AI Act, the GDPR and the standards that govern AI.

Browse all analyses & guides

Agentic AI: what changes when the system acts, not just decides

Analysis
Human oversight

Provider or deployer: which role are you under the EU AI Act?

Analysis
Accountability

Control-level compliance: the EU AI Act as an instrumented system

Analysis
Accountability

EU AI Act timeline 2025–2028: all deadlines after the omnibus agreement

Reference
Accountability

High-risk AI or not? classification guide for deployers

Guide
Accountability

First steps: EU AI Act compliance for deployers

Guide
Accountability

From free knowledge to deep mastery

A free, public knowledge base to understand AI governance, and a paid Academy to master it in depth.

Free

Responsible AI

Plain-language answers on the EU AI Act, the GDPR and the standards that govern AI, organized around the seven pillars of responsible AI. Guides, analyses and legal references, open to everyone.

Browse the knowledge base
Academy

AI Governance Academy

Deep, original lessons, scenario-based questions and honest readiness tracking. The first track prepares you for the AIGP exam; more frameworks follow.

See the Academy

Understand AI governance in five reads

  1. 1What is AI governance
  2. 2The seven pillars of responsible AI
  3. 3From principle to control
  4. 4AI risk management in practice
  5. 5The law and the frameworks
The seven pillars

The seven pillars of responsible AI

The EU AI Act, ISO/IEC 42001, the NIST AI RMF and the OECD AI Principles do not share one structure, but they address a recurring set of concerns: fairness, safety and reliability, privacy, security and robustness, transparency and explainability, accountability, and human oversight.

The seven pillars organize the knowledge base around those principles. GovCompass connects each principle to the harms it guards against, the risks those harms create, the controls that address them, and the evidence that shows the controls operate. The pillars structure the content. They do not replace the sources.

FairnessAI outcomes don’t discriminate, comparable people get comparable treatment.AI in recruitment: risks, bias and what the EU AI Act already requires
Safety & reliabilityThe system does what it is meant to, performs accurately, and fails safely.High-risk AI or not? classification guide for deployers
PrivacyPersonal data is processed lawfully and only as far as needed (GDPR).EU AI Act and GDPR: how do the two regulations relate?
Security & robustnessThe system resists attacks, manipulation and adversarial or unexpected input.The agentic threat surface, translated for AI Officers
Transparency & explainabilityPeople can understand how the system works and why it reached a decision.Art. 50 EU AI Act, transparency: inform users about AI interaction
AccountabilityClear ownership, documentation and evidence stand behind every AI decision.Provider or deployer: which role are you under the EU AI Act?
Human oversightA competent person can review, override and stop the system.Human oversight: keeping people in control of AI
The Academy

Master AI governance, the deliberate way

The AI Governance Academy turns the frameworks into applied skill: learn each concept, practice it on real scenarios, and get an honest measure of how ready you are.

Step 1

Learn

Short, stepped lessons: the mechanism, the why, a worked governance scenario and the exam traps, with a quick check at each concept.

Step 2

Practice

Scenario-based questions, graded on the server, with feedback that explains why the right answer is right and why each wrong one is wrong.

Step 3

Track honestly

A short diagnostic, a mastery map, and a readiness measure based on your question performance, not on how much you have read.

The method is deliberate: retrieval practice (recalling works better than re-reading) and interleaving (mixing topics to build the distinctions the exam tests). Timed mock exams and spaced review are planned. We will not claim they exist before they do.

What it covers

Four domains, mapped to the official blueprint and its question weighting, growing as content is published.

IFoundations of AI governanceWhat AI is, why it is hard to govern, and how an organization sets expectations, roles and policies for trustworthy AI across the life cycle.~21% of the exam· 3 competencies· 14 lessons
IILaws, standards and frameworks apply to AIHow existing data-privacy and other laws reach AI systems, the main elements of AI-specific laws (anchored on the EU AI Act), and the principal standards and frameworks (OECD, NIST AI RMF, ISO).~25% of the exam· 4 competencies· 10 lessons
IIIGovern AI developmentGovernance through the build: design decisions and risk management, data used for training and testing, and the controls around release, monitoring and maintenance.~27% of the exam· 3 competencies· 11 lessons
IVGovern AI deployment and useGovernance on the receiving end: deciding whether and how to deploy, assessing systems and vendors, and governing use, monitoring and downstream harms over time.~27% of the exam· 3 competencies· 11 lessons
Start studying Read a full lesson free

No exam outcome is ever guaranteed, this is serious preparation, honestly measured.