IT Glossary

AI Governance

AI governance decides which AI tools may be used, for what, with which data and under whose responsibility. Building blocks, frameworks and monitoring.

November 2, 2026

What is AI governance?

AI governance is the organization-wide framework of policies, roles and controls that decides which AI tools and models may be used, for which purposes, with which data and under whose responsibility, and how their use, risk and cost are monitored over time. It covers bought AI tools, AI features inside existing SaaS applications and models built in-house, not only autonomous agents.

Key takeaways

  • AI governance decides which AI tools may be used, for what, with which data and under whose responsibility, across the whole organization.
  • AI governance is broader than AI agent governance, which focuses on the identities, permissions and actions of autonomous agents.
  • The NIST AI Risk Management Framework (AI RMF 1.0), released on 26 January 2023, is a reference intended for voluntary use.
  • ISO/IEC 42001:2023 specifies requirements for an artificial intelligence management system, giving AI governance a structure that can be audited.

How AI governance works

  • Inventory: a register of the AI tools, AI features and models in use, including those adopted without IT approval, with their users and cost.
  • Approval path: a defined route to request a new AI tool, with security, legal and data checks before it is rolled out.
  • Data rules: which categories of data may go into which tools, for example no customer personal data in a tool without a data processing agreement.
  • Owners: a named business owner and a technical owner for each tool, accountable for how it is used and whether it is renewed.
  • Monitoring and review: usage, access and spend are tracked continuously, and the whole register is reviewed on a fixed cycle, such as every quarter.

AI governance or AI agent governance: where does each one stop?

  • AI governance: the organization-wide layer. It sets the rules for every AI use, from a chat assistant to an AI feature in the CRM, and records who decided, on what basis and with which data.
  • AI agent governance: a narrower layer for autonomous agents that act on systems. It deals with agent identities, delegated permissions, tokens and action logs.

A company can control two internal agents tightly and still not know which 30 AI tools its staff use. Published frameworks help structure the broader layer. The NIST AI Risk Management Framework, released on 26 January 2023 and intended for voluntary use, organizes the work into Govern, Map, Measure and Manage functions. ISO/IEC 42001:2023 specifies requirements for an artificial intelligence management system (AIMS), built like other ISO management system standards such as ISO/IEC 27001.

Examples and use cases

A 250-person fintech wants a usable AI policy before its next customer security review. The first inventory finds 19 AI tools, including the same writing assistant paid three times by different teams, and no named owner for most of them. The AI governance work then follows the building blocks: one approved tool per need, an owner for each, data rules written into the policy and a quarterly review of usage and spend.

Corma supports the inventory and monitoring parts: it discovers the AI tools in use, including shadow AI, records licenses, costs and renewals, tracks AI spend, and runs access reviews with audit evidence so that each quarterly review starts from current data.

Related concepts

FAQ

What does an AI governance framework include?

An inventory of AI tools, an approval path, data rules, named owners, monitoring of usage and cost, and a review cycle. Reference frameworks such as the NIST AI RMF and ISO/IEC 42001 help structure it.

Is the NIST AI Risk Management Framework mandatory?

No. The NIST AI RMF 1.0, released on 26 January 2023, is intended for voluntary use.

What is ISO/IEC 42001?

ISO/IEC 42001:2023 is the international standard that specifies requirements for an artificial intelligence management system (AIMS), in other words the processes an organization runs to govern its use of AI.

Who owns AI governance in a mid-market company?

Usually a small group: IT or security for tools and access, legal or the data protection officer for data rules, and finance for spend, with one person accountable for the AI register.

Request a demo