Identity governance is the missing control plane for IAM and AI

At a glance

What this is: This article argues that identity governance is the policy and oversight layer IAM lacks, because IAM authenticates and provisions access but does not continuously judge whether that access remains appropriate.

Why it matters: For IAM, IGA, PAM, and NHI practitioners, the distinction matters because operational access control without governance leaves toxic entitlements, segregation-of-duties conflicts, and audit exposure unresolved.

Context

Identity governance is the layer that decides who should have access to what, why, and for how long, while IAM handles authentication, session control, and provisioning. The article's central point is that enterprises can have functioning IAM and still fail on access risk if no governance layer continuously validates entitlements against business policy.

That gap becomes more visible across ERP, SaaS, databases, infrastructure, and AI initiatives. In practice, the issue is not whether users can log in, but whether access remains appropriate, defensible, and auditable once it exists.

Key questions

Q: What breaks when organisations rely on IAM without identity threat detection?

A: IAM can authorise access but cannot by itself show when access is being misused after it is granted. That leaves gaps around credential theft, token replay, privilege escalation, and lateral movement. Without ITDR, identity abuse can continue long enough to become persistence rather than a contained event.

Q: When does identity governance matter more than IAM controls?

A: It matters most when access decisions have business, compliance, or fraud consequences that go beyond login control. If the question is whether a role should exist, whether a user should keep it, or whether two entitlements should coexist, governance is the deciding layer.

Q: How can organisations tell whether access review is actually reducing risk?

A: Organisations should look beyond campaign completion and measure downstream removal, reopened exceptions, and the number of high-risk roles still present after review. If rejected access does not disappear from the target system, or if repeated cycles keep certifying the same excess entitlements, the programme is not reducing risk effectively.

Q: How should organisations govern AI agents alongside human identity and device access?

A: Organisations should treat AI agents as a separate identity class with their own entitlement boundaries, logging expectations, and approval model. Human IAM controls often assume interactive sign-in and review cycles, which do not fit autonomous or programmatic access. The safer approach is to define actor-specific policy and verify which access paths can be delegated without expanding trust unnecessarily.

Technical breakdown

IAM executes access, but governance decides whether access should exist

IAM is transactional. It provisions identities, authenticates sessions, and enforces the immediate allow or deny decision at login. Identity governance sits above that layer and asks whether the role, entitlement, or access path should exist at all, whether it creates toxic combinations, and whether it can be justified against policy and business context. That difference is why organisations can have SSO, MFA, and automated provisioning and still be unable to explain who can approve payments, create vendors, or post journals in core systems. Governance turns access from a static configuration into a controlled decision system.

Practical implication: Map each critical entitlement to an owner, policy, and review path instead of treating IAM provisioning as sufficient control.

Toxic access emerges when SoD and entitlement drift are left outside IAM

Segregation of duties, privilege creep, dormant accounts, and orphaned entitlements are governance problems, not pure authentication problems. IAM may create, change, and delete accounts, but it does not continuously evaluate whether a user now holds conflicting access across ERP, SaaS, or infrastructure. That is why audit failures often surface long after the identity was provisioned. The article's core mechanism is that coarse-grained roles can look correct at creation time while becoming risky through accumulated entitlement drift and hidden conflict combinations.

Practical implication: Continuously certify and reconcile entitlements across business systems, not just at onboarding or role assignment.

AI identities inherit the same governance gap as human users

As AI agents, service accounts, and machine identities become first-class actors, the same policy questions reappear in a new form: what can act, under what conditions, and for how long. Traditional IAM was built to authenticate identities and issue access, not to govern prompts, actions, and data access for autonomous or semi-autonomous systems. That means AI initiatives can inherit the same blind spots that already affect human access, but with faster execution and broader downstream impact. The governance problem is therefore not separate from IAM, it is an extension of the same control model into non-human execution paths.

Practical implication: Apply the same access policy, lifecycle, and audit controls to AI identities that you already expect for human identities.

Threat narrative

Attacker objective: The practical objective is to exploit legitimate but poorly governed access paths to reach business actions, sensitive data, or audit-relevant control failures.

1. Entry occurs through normal IAM provisioning, SSO, or automated account creation, which establishes valid access without testing whether the entitlement is actually appropriate.
2. Escalation happens when hidden privilege combinations, dormant accounts, or over-broad roles accumulate across ERP, SaaS, or infrastructure without governance review.
3. Impact follows when improper access enables fraud, misconfiguration, audit failure, or compliance weaknesses that IAM alone cannot detect or explain.

Breaches seen in the wild

• Azure Key Vault Contributor escalation 2024: Datadog found Azure Key Vault Contributor could add itself to access policies and read every secret, key and certificate in a vault.
• Dropbox Sign breach 2024: A compromised back-end service account gave attackers Dropbox Sign customer data, including API keys, OAuth tokens and MFA information.

Read our 52 NHI Breaches Analysis report for a comprehensive view of breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Identity governance is the missing control plane because IAM answers access execution, not access justification. IAM can authenticate users, provision accounts, and enforce a login decision, but it does not determine whether the entitlement should exist or remain in place. That leaves organisations with operational access and no policy authority above it. The implication is that identity programmes need a governance layer to decide, certify, and explain access rather than simply issuing it.

Toxic entitlement drift is the governance failure mode that most IAM stacks cannot see. Over-privileged roles, dormant accounts, and hidden segregation-of-duties conflicts often persist across ERP, SaaS, cloud, and infrastructure even when IAM is functioning normally. This is not an authentication failure. It is a lifecycle and policy failure in which access becomes riskier over time without revalidation. Practitioners should treat entitlement drift as a control-plane defect, not an administrative nuisance.

AI identities make the governance gap structurally more urgent. The article's most important extension is that AI agents, service accounts, and machine identities need the same lifecycle and policy discipline as human users. Traditional IAM was not designed to govern prompts, actions, and data access for non-human actors. The implication is that identity governance now has to span human and non-human execution, or risk multiplying the same blind spots at machine speed.

Access review programmes fail when they are treated as paperwork instead of decision infrastructure. Certifications that do not connect to business context, entitlement normalization, and remediation workflows simply document exposure after the fact. Governance earns its value when it becomes the system that proves who should keep access, why they keep it, and what changed in the meantime. Practitioners should measure governance by decision quality, not review volume.

Identity governance is becoming the field's shared control plane across IAM, PAM, and AI governance. The article correctly frames governance as the connective tissue that aligns identity decisions with risk, compliance, and evidence. That matters because enterprise identity is no longer just people at login. It is humans, service accounts, and AI actors moving through the same policy and audit expectations. Practitioners should design for one control plane, not separate silos.

From our research library:

• Nearly 60% of IT leaders cite restrictive cost and complexity as a weakness of legacy identity governance, according to the 2025 State of Identity Governance Report.

What this signals

Identity governance now functions as the control plane for access decisions that IAM alone cannot qualify. The practical shift is from provisioning-centric thinking to policy-centric governance, where access is continuously judged against business context, SoD, and evidence requirements. Teams that stop at authentication will keep missing the deeper control failure: valid access that is no longer defensible.

AI governance will increasingly inherit identity governance patterns rather than replace them. The article points to a future where AI agents are managed through the same lifecycle and accountability concepts used for human and machine identities. For practitioners, that means the hardest part is not adding another tool, but extending existing policy logic to non-human execution paths without creating a second governance silo.

For practitioners

• Map business-critical entitlements to governance owners Assign explicit policy owners for roles that can move money, change suppliers, post journals, or alter infrastructure so access decisions are reviewable against business risk, not just technical role names.
• Separate provisioning from entitlement approval Keep IAM responsible for account creation and authentication, but require an independent governance decision for who gets what access and why before the entitlement is activated.
• Build recurring certification around toxic access Target SoD conflicts, over-privileged roles, orphaned accounts, and dormant access for review because these are the patterns most likely to survive normal IAM operations.
• Extend governance to AI identities and service accounts Apply the same lifecycle, least-privilege, and audit requirements to AI agents, service accounts, and machine identities so non-human access is not exempt from policy.
• Use audit evidence as a control outcome Track whether governance can prove access justification, certification status, and remediation history for regulators and auditors instead of treating reporting as a separate exercise.

Key takeaways

• IAM can execute access decisions, but it cannot determine whether those decisions remain appropriate, which is why governance becomes the control layer that closes the gap.
• The article's core warning is that toxic access, SoD conflicts, and orphaned entitlements can persist even in organisations with mature SSO, MFA, and provisioning.
• Practitioners should treat identity governance as the mechanism for entitlement justification, certification, and evidence, including for AI identities and service accounts.

Key terms

• Identity Governance: Identity governance is the set of controls that defines who approves access, who owns it, how it is reviewed, and when it is removed. In practice, it turns identity management from a deployment task into a durable control system that can withstand audits, organisational change, and operational growth.
• Segregation of Duties: Segregation of Duties is a control principle that prevents one person or role from combining incompatible permissions that could create fraud, error, or undetected change. In ERP environments, it must account for roles, transactions, approvals, and compensating controls across business processes.
• Toxic Access Combination: A toxic access combination is a set of permissions that becomes dangerous when granted together, even if each entitlement looks acceptable on its own. In identity governance, these combinations matter because they can enable misuse, separation-of-duties failures, or broader compromise.
• Access Certification: Access certification is the periodic review of whether an identity still needs its current entitlements. For NHIs, certification is only reliable when reviewers know the identity's owner, purpose, and expiry, otherwise stale machine access can persist long after the original use case has ended.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity lifecycle are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are building or maturing an IAM programme, it is worth exploring.