Unmanaged identities are widening identity risk across human and NHI estates

At a glance

What this is: NHI Mgmt Group's analysis, drawing on Delinea's blog post: Identity sprawl across human and NHI estates creates security, compliance, and operational exposure when discovery and governance lag deployment.

Why it matters: IAM, PAM, and IGA teams need this because unmanaged human and non-human identities break inventory, lifecycle, and accountability assumptions across the full identity estate.

Context

Unmanaged identities are identities that exist outside normal identity management, meaning they are not tracked, governed, or protected by standard controls. The article frames this as a growing governance gap across workforce users, developers, IT admins, machine identities, and agentic AI, where the problem is not just visibility but control loss across the full identity chain.

The core IAM issue is that unmanaged identities create access paths that do not age cleanly through joiner-mover-leaver, recertification, or privilege review processes. When identities can appear through shadow IT, rapid cloud adoption, outsourced operations, or AI workflow sprawl, traditional lifecycle models stop describing the environment accurately.

Key questions

Q: What breaks when non-human identities are managed outside the IAM operating model?

A: What breaks is accountability. Without IAM ownership, non-human credentials drift into fragmented secrets tools, inconsistent review cycles, and orphaned access that persists after the workload changes. That is how machine identities become invisible trust dependencies.

Q: Why do non-employee identities create so much audit and compliance risk?

A: They often sit outside the normal HR-driven lifecycle and can change faster than manual governance can track. If approvals, entitlements, and offboarding are not linked to a single identity record, auditors cannot easily verify that access was justified, time-bounded, and removed on schedule. Compliance becomes a documentation problem because the control evidence is missing.

Q: How can security teams spot unmanaged identity problems early?

A: Look for identities with unclear ownership, shared privileged access, dormant accounts that still authenticate, and machine or agentic identities created outside approved workflows. Those signals usually appear before a breach as drift between what systems believe exists and what is actually operating in the environment.

Q: Should organisations treat agentic AI as a workload identity problem or an AI problem?

A: They should treat it as both, but the control failure shows up first in workload identity. The AI may decide what to do, but the blast radius is governed by service accounts, tokens, permissions, and orchestration rules. If those controls are weak, the agent’s autonomy simply moves faster than the guardrails.

Technical breakdown

How unmanaged identities escape identity inventories

An unmanaged identity is one that never enters the authoritative inventory, or enters it so late that it no longer reflects reality. That can happen through shadow IT, temporary access, outsourced administration, fast-moving development work, or machine and AI workloads created outside governance workflows. Once an identity is outside the inventory, downstream controls such as ownership, policy assignment, review, and revocation are no longer reliable because the system cannot govern what it cannot enumerate.

Practical implication: inventory coverage has to be treated as a control outcome, not a reporting exercise.

Why agentic AI intensifies unmanaged identity risk

Agentic AI identities are different from ordinary automation because they can make decisions, interact with other agents, and complete tasks independently. The article notes that these agents can adapt their behavior, spin up additional models or agents, and operate across cloud, on-premises, and hybrid environments, which fragments oversight. That means the identity boundary is not fixed at provisioning time, and the governance problem becomes one of runtime containment as much as access assignment.

Practical implication: identity governance for agentic AI must track runtime behaviour, not only initial provisioning.

Why access control breaks down once identities become unmanaged

Once an identity is unmanaged, the usual protections around least privilege, JIT access, and review cycles lose their anchoring point. The article ties unmanaged identities to credential theft, privilege escalation, lateral movement, and unauthorised access because permissions persist without clear ownership or retirement. In practice, the failure is structural: standing access remains available long after the operational reason for it has disappeared, and teams cannot confidently certify what they never fully discovered.

Practical implication: access governance needs continuous discovery before least privilege and recertification can work effectively.

NHI Mgmt Group analysis

Unmanaged identity is an inventory failure before it is an access failure. The central governance problem is not that identities are merely poorly secured, but that they fall outside the system of record that makes security decisions possible. Once that happens, every later control, from review to offboarding, becomes probabilistic rather than authoritative. The practitioner implication is that discovery coverage is the first governance control, not a supporting one.

Agentic AI turns unmanaged identity from a lifecycle issue into a runtime governance problem. These identities can alter behaviour, coordinate with other agents, and proliferate across environments faster than static review models can absorb. That means the old assumption that identity state remains stable long enough to be certified is breaking down. Practitioners need to treat agentic AI as a class of identities whose governance boundary is dynamic, not fixed.

Continuous discovery is now the only defensible answer to identity sprawl. The article is right to connect unmanaged identities to workforce, developer, machine, and AI estates because the control failure is shared even when the actor type differs. What changes across those estates is not the need for governance, but the speed and shape of drift. The implication is that IGA, PAM, and CIEM must work as one operating model rather than separate programs.

Least privilege loses value when ownership is unknown. Privilege reduction depends on knowing who or what the identity belongs to, why it exists, and when it should disappear. Unmanaged identities collapse those assumptions, so even a well-designed privilege model cannot reliably constrain access if the subject of the policy is invisible. The practitioner conclusion is that governance and privilege control fail together when lifecycle accountability is absent.

Identity blast radius is the right concept for unmanaged estates. The risk is not just more identities, but more uncontained identities that can amplify a single mistake into a broader security, compliance, or operations problem. That is especially true where shared privileged accounts, outsourced administration, or rapidly deployed machine identities blur responsibility. Teams should measure how far one unmanaged credential can travel before detection or revocation.

From our research library:

• 88% of organisations have embedded AI agents in their workflows, according to KPMG's 2026 report.
• Read next: Agentic AI Identity Maturity Model

What this signals

Identity blast radius: unmanaged identities increase the number of places an attacker can gain durable access, but the deeper issue is that control boundaries no longer match operational reality. Security teams should expect the weakest part of the programme to be discovery coverage, not just privilege policy.

Agentic AI makes this worse because access can proliferate faster than review cycles can absorb. A control model that assumes identities remain stable between certification points will miss the class of identities that appear, act, and change too quickly to be governed manually.

Only 13% of organisations feel extremely prepared for the reality of agentic AI despite the majority racing toward autonomous adoption, according to the 2026 Infrastructure Identity Survey. That gap means governance for unmanaged identities is becoming a present-tense operational requirement, not a future planning exercise.

For practitioners

• Map unmanaged identity sources across all estates Build a single discovery workflow for workforce, developer, admin, machine, and AI identities across cloud, on-premises, and hybrid platforms. Prioritise sources where shadow IT, outsourced operations, and short-term access create identities outside the identity system of record.
• Automate joiner-mover-leaver handling for non-human identities Apply lifecycle controls to service accounts, machine identities, and agentic AI identities so that creation, change, and offboarding are not left to manual follow-up. Tie ownership to each identity so deprovisioning can happen when the underlying workload, project, or vendor relationship ends.
• Enforce least privilege and JIT for every identity class Remove persistent permissions from identities that do not need them and issue access only when the task requires it. This matters as much for workloads and agents as it does for people because unmanaged privilege is what turns discovery gaps into breach paths.
• Centralise access reviews and anomaly detection Use periodic certification to catch stale access, then pair it with automated anomaly detection so reviews are not the only line of defence. Focus on orphaned accounts, shared privileged access, and identities that appear in one platform but not another.
• Unify PAM, IGA, and CIEM operating data Correlate privileged access, governance records, and cloud entitlements so unmanaged identities do not hide inside different administrative planes. The objective is to close the gap between what exists technically and what is still governed operationally.

Key takeaways

• Unmanaged identities are a governance problem as much as a security problem, because once identities fall outside the inventory, normal control points no longer apply.
• The article links identity sprawl to security, compliance, and operational exposure across human, machine, developer, and agentic AI estates.
• The practical response is continuous discovery, lifecycle automation, and unified oversight across IGA, PAM, and CIEM.

Key terms

• Unmanaged identity: Any identity that can authenticate or act without being fully controlled by the organisation’s standard identity stack. That includes service accounts, API keys, tokens, and some AI agents. The risk is not just visibility loss. It is loss of ownership, lifecycle control, and reliable revocation.
• Identity Sprawl: Identity sprawl is the uncontrolled growth of identities, entitlements, and credentials across an environment. For NHIs, it usually appears when automation creates accounts faster than governance teams can inventory, review, and remove them. The result is hidden access, weak accountability, and a wider attack surface.
• Agentic AI Identity: The complete set of credentials, permissions, and governance controls applied to an autonomous AI agent, covering authentication, authorisation, action logging, and access revocation. Distinct from traditional NHI because agent identities are often ephemeral, delegated, and multi-hop.
• Identity Blast Radius: The amount of damage a compromised identity can cause across systems, data, and infrastructure. In NHI environments, it is shaped by permissions, network reach, and administrative capability rather than by the credential alone. Reducing blast radius is a containment strategy that limits lateral movement and data exposure.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity security 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.