Where to start with AI governance: inventory before policy

← Alle Beiträge
Künstliche Intelligenz · EN

Where to start with AI governance: inventory before policy

If an organization is just beginning to set up AI governance, what’s the most important first step toward an effective AI risk management system without disrupting daily business?

My answer: Start with a structured inventory.

Before any framework, policy, or control becomes meaningful, you need visibility on what is actually in use:

• Which use cases are running across the business?
• Which tools (including AI features embedded in existing software) are deployed?
• Which data flows into and out of these systems?

Only on this basis can you:

Apply a risk-based classification along the lines of the EU AI Act (prohibited / high-risk / limited / minimal risk).
Define a pragmatic traffic-light system for approval pathways — green for low-risk, yellow for conditional use, red for cases requiring deeper assessment (DPIA, FRIA, security review).
Keep daily business running while governance matures in parallel.

Two references worth consulting:

1) NIST AI RMF 1.0: the framework explicitly recommends that, after establishing governance, organizations begin with the MAP function, whose purpose is to establish the context to frame risks related to an AI system and to map risks and benefits for all components of the AI system including third-party software and data.

2) ISO/IEC 42001:2023 (AI Management System): Annex A defines controls across nine objectives. The standard operationalizes exactly the kind of structured, auditable approach that turns an inventory into a governance system.

Zu diesem Beitrag gibt es auf LinkedIn ein Dokument als Anhang. Auf LinkedIn ansehen →

Auf LinkedIn diskutieren