Modeling the AI agent explosion and its identity governance burden

At a glance

What this is: This is an analysis of how AI agent proliferation changes identity governance, with the key finding that agent identities may soon outnumber human users by 25:1.

Why it matters: It matters because IAM, IGA, PAM, and NHI programmes will have to govern far more short-lived identities than current human-centric processes were built to handle.

By the numbers:

• AI identities will soon outnumber human users 25:1.
• A 1,000-person company could suddenly be managing hundreds of thousands or even millions of AI agent identities annually.

👉 Read C1.ai's analysis of the AI agent identity explosion

Context

AI agent identity is becoming a density problem, not just an access problem. As organisations attach agents to business applications and employee workflows, the number of non-human identities can rise far faster than the teams responsible for governing them.

That shift breaks the operating assumptions behind manual access reviews, static provisioning, and slow lifecycle processes. For identity teams, the practical question is how to govern short-lived machine and agent identities at enterprise scale without losing visibility or control.

Key questions

Q: How should security teams govern AI agents that use multiple identity layers?

A: Security teams should inventory every identity layer an agent can use, including static credentials, session identities, embedded tool identities, and any delegated relationships between agents. Governance fails when one layer is controlled while another remains open, because the agent can still act through the weaker path. Treat the layered identity surface as the actual access boundary.

Q: Why do AI agents make non-human identity governance harder?

A: AI agents make governance harder because they can request tools, act autonomously, and change behaviour across sessions while still relying on machine credentials. That increases the number of access paths security teams must supervise. The result is a stronger need for task-scoped access, explicit ownership, and continuous monitoring of what the agent can reach.

Q: What breaks when organisations treat agent identities like service accounts?

A: What breaks is accountability. Service accounts are usually governed as rigid, task-bound identities, but agents can inherit authority, choose actions, and continue moving through systems at runtime. Treating them the same way hides delegation context and makes it harder to explain why an action was allowed.

Q: How do organisations keep AI agent sprawl under control?

A: Organisations need a clear inventory, a named owner for each workflow, and automatic expiry for each agent credential. They also need telemetry that shows when agents are accessing data or invoking tools outside their expected scope. Without those controls, sprawl becomes a governance problem rather than an automation gain.

Technical breakdown

Why agent density breaks traditional identity models

Traditional IAM assumes identities are countable, relatively stable, and governable through periodic review. AI agents invert that model because each task, workflow, or app integration may create a new identity instance with its own credentials, scope, and lifecycle. At scale, the challenge is not only authentication but inventory, ownership, and revocation across a constantly changing population. The result is identity sprawl with a much shorter operational half-life than human or service accounts.

Practical implication: replace periodic, manual governance with automated discovery, classification, and expiry for agent identities.

Why ephemeral credentials matter more than static access

The article’s core technical point is that agent lifetimes are compressing from months or years to minutes. That means credentials, tokens, and delegated permissions need to be provisioned just in time and revoked at the end of the task, not left in place for the next run. In NHI terms, the threat is standing privilege persisting longer than the work itself. In agentic environments, the control objective shifts from durable access management to continuous issuance and teardown.

Practical implication: design agent access around short-lived credentials, explicit expiry, and deterministic teardown.

Why AI-native governance is becoming necessary

The article points to a scale problem that conventional identity platforms struggle to absorb. Once agents begin to spawn other agents, the governance burden compounds across provisioning, audit, and policy enforcement. That is why AI-native identity models matter: they can express machine-to-machine context, runtime scope, and delegated action in ways traditional human IAM workflows do not. This is less about replacing IAM than extending it to identities that behave like workloads and move at software speed.

Practical implication: build governance workflows that can track agent ownership, runtime scope, and chained delegation in real time.

NHI Mgmt Group analysis

AI agent density is now an identity governance problem, not an application feature problem. Once every business workflow can spawn agents, the number of identities becomes operationally meaningful in its own right. The article’s 25:1 projection is less important than the direction of travel: governance demand grows faster than human staffing models can absorb. Practitioners should treat agent population growth as a control design input, not a deployment side effect.

The access review model assumes identities persist long enough to be reviewed, and that assumption collapses for agents. The article explicitly describes agent lifecycles measured in minutes, not years. That means certification cycles, recertification queues, and periodic access attestations are built for a slower object than the one now appearing in the enterprise. The implication is not merely more automation, but a rethinking of what can meaningfully be reviewed after execution has already ended.

Ephemeral credentials create identity blast radius if ownership and expiry are not tied to the task. When agents are spawned at scale, leftover permissions and unclear accountability become the primary failure modes. A named concept here is identity density shock: the point at which identity volume grows faster than governance primitives designed for humans and long-lived service accounts. Practitioners need to recognise that density itself is now a security variable.

AI agents extend NHI governance into runtime behaviour, which is where traditional IAM becomes weakest. The article points toward a world where access is generated dynamically, multiplied by workflow, and consumed by software rather than people. That pushes identity strategy from static entitlement management toward runtime lifecycle control, especially where agents can query data, trigger actions, and chain tasks across systems. Identity teams should assume the control surface has moved into execution time.

Human staffing ratios are the wrong planning unit for agent governance. The article correctly rejects the idea that organisations will hire proportionally more staff to manage agent identities. That matters because the programme model, not just the tooling, has to change. If governance capacity scales linearly while identity density scales exponentially, every access control and lifecycle process will degrade under load. Practitioners should plan for machine-scale identity administration, not human-scale exception handling.

From our research:

• 92% agree governing AI agents is critical to enterprise security, yet only 44% have implemented any policies to do so, according to AI Agents: The New Attack Surface report.
• Only 52% of companies can track and audit the data their AI agents access, leaving 48% with a complete blind spot for compliance and breach investigation.
• That governance gap pairs with the Ultimate Guide to NHIs , Lifecycle Processes for Managing NHIs for the operational model and the OWASP Agentic AI Top 10 for the threat surface.

What this signals

Identity density shock: once AI agent populations scale faster than governance staffing, the control problem shifts from access approval to continuous lifecycle enforcement. Teams should expect their IAM and IGA backlogs to become operationally irrelevant unless agent discovery, ownership, and expiry are automated.

With 71% of organisations already reporting that NHIs are not rotated within recommended time frames, the same lifecycle weakness will show up quickly if agent credentials are allowed to persist. The right response is to align agent governance with existing NHI lifecycle discipline, not to bolt it onto human access processes.

Practitioners should watch for agent sprawl in the same places they watch for shadow IT: unmanaged app integrations, unowned workflows, and credentials with no clear offboarding path. The issue is not just volume, but the speed at which identity objects can appear and disappear before manual governance ever sees them.

For practitioners

• Inventory agent identities by business function Map every known AI agent to an owning team, application, and access scope before deployment expands further. Treat the inventory as a living control surface, not a one-time discovery exercise.
• Design for short-lived credentials and expiry Issue credentials only for the duration of the task, and require automatic teardown when the agent completes its work. Align this with the lifecycle model used for high-risk non-human identities.
• Bind approval and accountability to the workflow owner Make one business owner responsible for each agent-driven workflow, including periodic validation of scope, logs, and revocation paths. That reduces orphaned access when agents are created by automation.
• Expand identity governance controls to runtime events Trigger review, alerting, and revocation on observed agent behaviour such as new data access, tool use, or delegation changes. Static recertification alone will not keep pace with agent lifecycles measured in minutes.
• Use NHI lifecycle controls for agent populations Apply the same discipline used for service accounts and API keys to agent identities, especially around provisioning, rotation, and offboarding. The Ultimate Guide to NHIs is a useful reference point for structuring that lifecycle.

Key takeaways

• AI agent growth turns identity volume into a governance constraint, because manual IAM processes do not scale with ephemeral software-driven identities.
• The 25:1 ratio is a planning signal, but the deeper issue is that agent lifecycles measured in minutes invalidate review-based control models.
• Practitioners need ownership, expiry, and runtime visibility now, or agent sprawl will outrun both policy and staffing.

Key terms

• AI Agent Identity: The digital identity used by an autonomous AI agent to authenticate to external systems, APIs, and services. Managing AI agent identities is an emerging and rapidly evolving area of NHI security.
• Identity Density: Identity density is the rate at which identities accumulate relative to the organisation’s ability to govern them. For AI agents and other non-human identities, density becomes a security variable because high-volume, short-lived accounts can overwhelm manual lifecycle, review, and revocation processes.
• Ephemeral Credentials: Ephemeral credentials are short-lived access artefacts issued for a limited task or session. They reduce the window for abuse, but they only improve security when paired with strong scope limits, telemetry, and automatic revocation at task completion.
• Agent Sprawl: Agent sprawl is the uncontrolled growth of AI agents, scripts, and automation identities across teams and environments. It creates governance strain because each agent can introduce its own permissions, secrets, and ownership gaps, making revocation, review, and accountability harder to sustain.

What's in the full article

C1.ai's full blog covers the operational detail this post intentionally leaves for the source:

• The article’s reasoning model for comparing servers, VMs, containers, serverless, and AI agents across identity density.
• The specific examples of how a 1,000-person organisation can accumulate hundreds of thousands or millions of agent identities.
• The source’s own framing of ephemeral-by-default access and why short-lived credentials become essential.
• The call to action around AI-native IDP design and density-driven identity planning.

👉 C1.ai's full blog covers the density model, lifecycle implications, and planning assumptions in more detail.

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.