Skip to content
Innopulse Consulting
EU AI Act

What is AI governance?

Short definition

AI governance is the organisational framework through which a company steers its use of AI: responsibilities, policies, an inventory of systems, risk classification, approval of new applications, monitoring and documentation. It brings the requirements of the AI Act, data protection, information security and supervision together in one process.

Many companies already use AI widely, often without anyone having an overview. AI governance creates that overview and makes use steerable. It is not a bureaucracy project but the prerequisite for using AI quickly and without nasty surprises.

The building blocks

First, responsibilities: who decides on new AI applications, who owns individual systems, who assesses risks? A RACI matrix clarifies this. Second, a policy setting out which tools may be used with which data. Third, an AI inventory of all systems with their role and risk class. Fourth, an approval process for new applications. Fifth, monitoring and regular review in operation.

What feeds into it

Depending on role and risk class, the AI Act requires AI literacy measures, transparency, human oversight and, for high-risk systems, comprehensive risk management. Data protection law requires legal bases, contracts with vendors and impact assessments where needed. For Swiss financial institutions, FINMA set out expectations for governance and risk management in AI use in Guidance 08/2024. And ISO/IEC 42001 offers a certifiable standard for AI management systems. Good governance brings these requirements together in one process rather than building a separate structure for each.

Proportionate, not maximal

An SME does not need an AI committee meeting monthly. It needs a responsible person, a short policy, a current inventory and a clear rule on who approves new tools. A financial institution or a provider of high-risk systems needs more: documented risk processes, independent review and ongoing monitoring. Depth follows risk.

Common mistakes

A policy that bans everything and is therefore ignored. An inventory created once and never maintained. Governance only for in-house developments, while AI features bought into existing software go unnoticed. And approval processes so slow that teams bypass them.

Getting started

Start with inventory and policy; the AI usage policy checklist and the 30-day AI Act readiness plan describe the steps. We support the build-up through our AI governance set-up offer.

Sources and references
  • FINMA Guidance 08/2024: governance and risk management when using artificial intelligence
  • ISO/IEC 42001:2023, artificial intelligence management systems

Legal position as of October 2026. Not legal advice for an individual case.

EU AI Act is our specialty

Innopulse doesn't just explain terms — we put them into practice for DACH companies.