Who should own AI identity governance in an enterprise programme?

Why This Matters for Security Teams

AI identity governance fails when it is treated as a narrow IAM task instead of a shared control plane across discovery, approval, access scope, and monitoring. The ownership question matters because AI agents, service accounts, tokens, and pipelines can all create machine-speed privilege paths that traditional review cadences miss. NHI Management Group’s Top 10 NHI Issues and NIST Cybersecurity Framework 2.0 both point to the same operational reality: identity governance must connect policy, inventory, and enforcement.

The enterprise risk is not just excess privilege. It is fragmented accountability. IAM may own authentication standards, IGA may own approvals, PAM may own elevation, and security architecture may own control design, but no single team can safely own the full lifecycle alone. Current guidance suggests the programme owner should be the function that can coordinate all four without slowing delivery to a standstill. In practice, many security teams encounter shadow AI identities only after a token, connector, or agent has already been granted broader access than intended.

How It Works in Practice

The most effective operating model is a federated one. IAM provides the identity substrate, IGA governs lifecycle and recertification, PAM handles privileged elevation and vaulting, and security architecture sets the target control pattern for agents and other NHIs. The programme owner, however, should be accountable for making those functions work as one system. That means defining control objectives, ownership boundaries, escalation paths, and evidence requirements before AI deployment expands.

For autonomous and semi-autonomous AI workloads, ownership has to reflect how the identity is actually used. A service principal running a batch job is not governed the same way as an AI agent that can choose tools, chain actions, and request new permissions mid-task. NHI Management Group’s Ultimate Guide to NHIs — Lifecycle Processes for Managing NHIs is useful here because it frames identity as a lifecycle, not a one-time approval. In parallel, the NIST Cyber AI Profile (IR 8596) reinforces the need for governance, measurement, and operational monitoring across AI use.

• Use IAM to define identity types, trust boundaries, and machine authentication standards.
• Use IGA to register every AI identity, owner, purpose, and approval path.
• Use PAM to control elevation, break-glass access, and just-in-time privilege.
• Use security architecture to enforce policy patterns for APIs, tools, and agent runtime access.
• Use continuous monitoring to detect dormant, over-scoped, or orphaned AI identities.

In practice, the programme should also require a clear business owner for each AI system, because technical ownership without operational accountability quickly degrades into unmanaged exceptions. These controls tend to break down when AI teams can create tokens, connectors, or agents outside the central identity workflow because review and revocation happen too slowly.

Common Variations and Edge Cases

Tighter ownership often increases coordination overhead, requiring organisations to balance speed against control maturity. That tradeoff is real, especially in product-led environments where development teams want immediate access for testing and deployment. Best practice is evolving, but there is no universal standard for whether IAM or IGA should be the formal programme lead; what matters is that one accountable function can force cross-team decisions and evidence collection.

Some enterprises split ownership by environment. For example, platform engineering may own non-production AI identities, while security owns production approval gates and monitoring. That can work if the handoff is explicit and if exceptions are tracked centrally. It becomes risky when agents can move from test to production with the same credentials, or when contractors, vendors, and embedded tools create hidden identity sprawl. The 52 NHI Breaches Analysis shows why lifecycle visibility matters more than organisational charts, and the Ultimate Guide to NHIs — Regulatory and Audit Perspectives is a practical reminder that auditors will ask who owned the control, not just who configured it. The hardest cases are mergers, federated cloud estates, and agentic AI programmes where identity ownership is split across business units and no single team can enforce a common standard.

Related resources from NHI Mgmt Group

• Who should own AI identity governance in an enterprise IAM programme?
• Who should own AI agent governance in an enterprise identity programme?
• Who should own identity and data exposure decisions in a governance programme?
• Why is single-provider AI agent governance not enough for enterprise security?