TL;DR: Non-human identities are moving past 100:1 over the next two years, and Britive argues that vault-centric PAM cannot scale into that ratio because agentic identities are short-lived, runtime-selected, and often instantiated on demand. The core problem is not volume alone: it is the collapse of the standing-credential assumption that legacy NHI governance was built on.
At a glance
What this is: Britive argues that vault-centric PAM cannot keep up as non-human identities cross 100:1, because agentic identities are ephemeral and created at runtime.
Why it matters: IAM and PAM teams need to treat agentic identities as a different governance population, because design assumptions built for durable credentials do not hold when access is instantiated on demand.
Context
Vault-centric PAM assumes an identity lives long enough to receive, store, and rotate a credential. That model works for persistent service accounts and integration users, but it starts to fail when identities are created at runtime and disappear almost immediately after use.
The governance problem is not simply more inventory. It is that agentic identities change the access pattern itself, which means lifecycle, authorization, and audit controls must shift from stored credentials to per-call decisioning.
For IAM leaders, the key question is no longer how to vault more identities. It is how to govern a population whose existence may last only seconds and whose tool use is determined at execution time.
Key questions
Q: What breaks when organisations try to manage agentic AI with vault-centric PAM?
A: Vault-centric PAM assumes an identity persists long enough for credential issuance, storage, and reuse to make sense. Agentic identities may exist only for a short task window and choose tools at runtime, so the reusable secret becomes the control failure rather than the control point. The access decision has to move into execution time, not sit in a vault.
Q: Why does per-call policy evaluation reduce risk for ephemeral agents?
A: Per-call policy evaluation shifts trust to the exact moment an agent tries to act, so access is granted only for a specific request and resource context. That reduces reliance on reusable credentials and aligns governance with the agent's actual runtime behaviour rather than a provisioned assumption.
Q: What signs show that an NHI programme is still built around standing access?
A: The clearest sign is when teams can inventory identities and rotate secrets, but cannot explain how short-lived agents are authorised at the moment of tool use. If ownership and audit depend on post-provisioning review, the programme is still assuming access persists long enough to be managed later.
Q: How should IAM teams govern agents that are instantiated by users on demand?
A: Treat them as a separate governance population with explicit runtime authorization, scoped tool access, and task-bound expiry. Do not let individual projects invent their own model, because inconsistent delegation quickly creates audit gaps and uneven blast radius across the estate.
Technical breakdown
Why standing credentials stop making sense for agentic identities
Vault-centric PAM was built around durable identities that can be enrolled, stored, and rotated over time. Agentic identities are different: they are short-lived, often created on demand, and select tools at runtime rather than following a fixed script. That combination collapses the usual vault workflow, because the credential may outlive the task by only seconds, if it exists at all. The result is a control model that keeps protecting something that is no longer the main security boundary. The real boundary shifts to the moment of authorization, not credential custody.
Practical implication: move governance attention from stored credentials to runtime authorization for ephemeral identities.
Per-call policy evaluation versus vault-based credential storage
Per-call policy evaluation checks the specific request, resource, and context at the moment an agent tries to act. That is a different control shape from vaulting, which assumes the credential is the object to secure between uses. In the article's model, this is the architectural pivot that supports Zero Standing Privilege for agentic identities. It does not eliminate governance work, but it changes where trust is established: at invocation time, not during long-lived credential handling. This is the key distinction practitioners miss when they try to extend vault-centric PAM by simply adding more inventory or rotation logic.
Practical implication: define approval and policy checkpoints at invocation time for high-value agent actions.
Why identity inventory alone will not govern the second wave
The first wave of NHIs was mostly persistent service accounts, CI/CD identities, and SaaS integration users. Those can be inventoried, assigned owners, and brought under lifecycle control. The second wave, driven by agentic systems, behaves differently because one user can instantiate many temporary agents in a day, each scoped to a task. That means inventory still matters, but it no longer solves the core governance problem. The issue is not just how many identities exist. It is whether the governance model can handle identities whose lifecycle, tool set, and access scope are decided at runtime.
Practical implication: redesign inventory, ownership, and audit processes for short-lived identities, not just large volumes.
Threat narrative
Attacker objective: The objective is to obtain usable access through a control model that still treats ephemeral agent actions as if they were durable credentials.
- Entry occurs when a user request instantiates an agentic identity on demand, rather than through pre-provisioned lifecycle management.
- Escalation occurs when the agent selects tools and resources at runtime, creating access scope that was not fully known at provisioning time.
- Impact occurs when standing-credential assumptions leave a persistent security surface around identities that should have been ephemeral.
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.
Read our 52 NHI Breaches Analysis report for a comprehensive view of breaches impacting Non-Human Identities including AI Agents.
NHI Mgmt Group analysis
Ephemeral access breaks the standing-credential premise: Vault-centric PAM was designed for identities that persist long enough to justify storing and rotating secrets. That assumption fails when the actor is an agentic identity because access is created at runtime, used briefly, and discarded before a vault workflow can add value. The implication is not just larger scale. It is that the governance object has changed, and the old lifecycle logic no longer describes it.
Zero Standing Privilege becomes the correct control shape for agentic populations: The article's strongest point is architectural, not operational. If access is granted per call and expires after the action completes, the security model follows the actor's actual lifecycle instead of forcing the actor into a vault-centric one. For practitioners, the conclusion is that standing credentials should stop being the default abstraction for agents.
Inventory is necessary but no longer sufficient: The first-wave NHI model worked because identities were stable enough to enumerate, assign, and review. Agentic identities can be multiplied by user action, making population growth inseparable from runtime behaviour. That means governance must move upstream into request-time policy, because post-hoc ownership and rotation do not govern identities that barely persist.
Agentic NHIs are a different population, not a larger one: This is the named concept that matters. The industry often talks about NHI growth as if the same controls can be stretched to cover more volume, but the article shows that the second wave changes lifecycle, timing, and accountability all at once. That means IAM programmes should reclassify the problem before they scale the wrong control plane further.
Standing-credential governance creates blast radius by design: When credentials exist between uses, they create a reusable attack surface even if the underlying workload is ephemeral. The article makes clear that adding agent vaulting on top of the old model preserves the problem rather than solving it. Practitioners should treat this as a control-design failure, not a capacity issue.
From our research library:
- NHIs outnumber human identities by 25x to 50x in modern enterprises, according to the Ultimate Guide to NHIs.
- 59% of organisations say they lack viable alternatives to standing privileged access for NHIs and AI agents, according to Delinea research.
- Read next: Ultimate Guide to NHIs — Key Challenges and Risks
What this signals
Agentic NHIs are a different population: Governance programmes that treat agentic systems as simply more NHI volume will overfit the wrong control model. The real shift is in lifecycle and timing, because the identity may appear only when a user asks for it and vanish after the action completes.
Vault-centric PAM keeps the wrong thing in focus. The control problem moves from storing credentials safely to deciding whether a request should be allowed at the moment of execution, which is why per-call authorization matters more than vault expansion.
A useful test for IAM teams is whether they can explain who owns a temporary agent, what authority it inherits, and how its access disappears when the task ends. If those answers are unclear, governance is still anchored to durable identities, not ephemeral ones.
For practitioners
- Inventory agentic identities already in use Include copilots, embedded SaaS assistants, and other runtime-created agents that teams can stand up without a formal procurement path.
- Classify the agent identity model explicitly Decide whether each agent inherits user policy, binds to a distinct identity, or acts through a delegated service account before projects diverge.
- Move high-value use cases to per-call evaluation Prioritise agents touching support cases, CRM data, security tooling, and finance workflows where runtime authorization can replace standing access.
- Rework governance metrics for ephemeral access Measure request-time authorization coverage, tool-scope decisions, and task-scoped expiry instead of counting only stored credentials.
Key takeaways
- The article argues that the problem is not simply more NHI volume, but a shift to short-lived agentic identities that old PAM models were not designed to govern.
- Its strongest evidence is architectural: access that is created on demand and scoped at runtime cannot be managed effectively as a standing credential.
- Practitioners should respond by moving governance from vault custody to per-call authorization and task-scoped expiry.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Non-Human Identity Top 10 and MITRE ATT&CK address the attack and risk surface, while NIST SP 800-53 Rev 5 and NIST CSF 2.0 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Non-Human Identity Top 10 | NHI-05 — Overprivileged NHI | The article centres on governance for agentic NHIs whose access outgrows vault-centric assumptions. |
| NHI-07 — Long-Lived Secrets | Vault-centric PAM assumes credentials persist long enough to manage, which this article says breaks for agents. | |
| Recommendation — Apply NHI-05 to replace standing access with tightly scoped runtime authorization for agentic identities. Eliminate long-lived secrets for ephemeral agents and move access decisions to task time. | ||
| NIST SP 800-53 Rev 5 | IA-5 — Authenticator Management | Credential lifecycle management is central to the article's critique of standing secrets for NHIs. |
| Recommendation — Use IA-5 to govern NHI authenticator issuance, rotation, and revocation against runtime use patterns. | ||
| NIST CSF 2.0 | PR.AA-05 — Access Permissions, Entitlements and Authorizations | The article is fundamentally about how authorization must shift from vault storage to request-time decisions. |
| Recommendation — Map agent access to PR.AA-05 and verify entitlements at the moment each tool call is made. | ||
| MITRE ATT&CK | TA0006;TA0008 — Credential Access; Lateral Movement | Standing credentials expand the attack surface for credential abuse and movement through connected resources. |
| Recommendation — Map reusable agent credentials to TA0006 and TA0008 to prioritise blast-radius reduction. | ||
Key terms
- Dynamic Ephemeral Identity: Dynamic Ephemeral Identity is a model in which credentials or authority exist only for a short operational window and are generated at runtime. It reduces the value of exposed secrets, but only if the environment can also limit what the identity is allowed to do while active.
- Standing Credential: A standing credential is any secret that remains usable until it is manually rotated or revoked. In NHI governance, it creates durable access that can be stolen, replayed, or propagated from trusted tooling unless runtime boundaries and expiry are built in.
- Per-Call Policy Evaluation: Per-call policy evaluation means checking whether a specific request is allowed at the exact moment it is made. For non-human access, this moves authorisation into the execution path and reduces reliance on durable secrets that can be stolen, reused, or mis-scoped.
- Zero Standing Privilege: A control model in which an identity does not keep persistent access unless it is actively needed. For NHIs, this means credentials and permissions are issued for a narrow task and then removed. It reduces the time window and reuse value of stolen access.
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.
Published by the NHIMG editorial team on August 15, 2026.
Updated on October 7, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org