AI agent identity sprawl is exposing IAM gaps in enterprise access

At a glance

What this is: This is a summary of Omdia-backed analysis on how AI agent identity sprawl is exposing gaps in enterprise IAM, especially around credential governance and non-human identity lifecycle control.

Why it matters: It matters because IAM teams now have to govern AI agents as non-human identities with real access, real credentials, and no human-style authentication or review cycle to rely on.

Context

AI agent identity sprawl is the governance problem here: the number of identities is not the only issue, but the way those identities are created, granted access, and left to operate beyond human-paced control loops. In practical terms, existing IAM models were built for people and stable service accounts, not for semi-autonomous software that requests access across many systems.

Omdia's report, as summarized by 1Password, argues that AI agents can generate dozens of non-human identities whenever authentication is needed and that traditional IAM was not designed to provision, de-provision, and govern them safely. That creates a direct connection between agentic AI adoption and the same lifecycle failures security teams already struggle with in NHI programmes, only at higher speed and wider scope.

The article's core point is that agentic AI changes the access problem from static entitlement management to continuous identity sprawl management. The governance question is no longer whether an agent can sign in once, but whether access can remain time-bound, auditable, and revocable while the agent is actively chaining tasks across business systems.

Key questions

Q: What breaks when AI agents rely on static secrets?

A: Static secrets break the trust model because they are reusable, portable, and often broader than the task requires. In an agent workflow, that can expose more systems than intended and make it difficult to prove why access was granted. Short-lived, brokered credentials are a better fit for runtime decision-making.

Q: Why do AI agents create more IAM risk than ordinary developer tools?

A: AI agents can make independent tool calls, chain actions, and authenticate with non-human identities while executing a task. Ordinary developer tools generally do not decide what to do next. That autonomy increases the chance of unintended access, makes attribution harder, and raises the value of session-level controls that show both intent and outcome.

Q: How should security teams govern AI agents that were never formally provisioned?

A: Security teams should govern them through runtime discovery, ownership mapping, and behavioural controls rather than relying only on directory records. If an agent appears in production traffic but not in identity systems, it still needs a documented owner, a known purpose, and an enforced access boundary. Without those, the organisation cannot prove accountability or lifecycle control.

Q: When does AI agent access need additional accountability controls?

A: Additional accountability is needed when an agent can touch sensitive business data, financial workflows, or shared enterprise applications without a human in the loop. In those cases, organisations need logs, approval boundaries, and lifecycle ownership that make the agent's actions explainable after the fact.

Technical breakdown

Why AI agents create non-human identity sprawl

AI agents do not behave like traditional application integrations. They can initiate multiple tool calls, request access as needed, and create additional identities whenever a workflow requires authentication. That means the identity footprint can expand dynamically during runtime instead of remaining fixed at provisioning time. In this model, every new application connection, API key, or delegated credential becomes part of the agent's operating surface. The governance challenge is not just authentication, but identity multiplicity: one agentic workflow can generate several NHI instances across different systems, each with its own access scope, lifecycle risk, and revocation requirement.

Practical implication: Map every agentic workflow to the identities it creates or consumes, not just to the primary account that launched it.

Why static credentials break agentic AI security

Static credentials are a poor fit for agentic AI because they persist longer than the task that needs them and can be reused outside the intended context. The article highlights risks such as hardcoded credentials, plaintext exposure, and unrestricted access to sensitive data, all of which expand blast radius when an agent operates continuously. In NHI terms, this is a lifecycle failure: the credential outlives the action, so the control boundary moves from runtime authorisation to post hoc cleanup. For AI agents, that is too late. The access pattern needs to be auditable and bounded at issuance, not assumed safe because it is embedded in software.

Practical implication: Replace hardcoded or long-lived agent credentials with time-bound issuance, scoped permissions, and revocation paths that work at machine speed.

Why traditional IAM does not fit agentic AI governance

Traditional IAM assumes relatively stable subjects, human-style identity proofing, and review cycles that can validate access over time. AI agents break those assumptions by operating continuously, requesting broad permissions, and lacking human authentication methods such as biometrics. That creates a governance gap between identity verification, permission management, and secure credential storage. The article also shows that the issue is not only access volume but auditability: if the agent can act, delegate, or trigger workflows without a clear event trail, IAM loses one of its core enforcement and accountability mechanisms.

Practical implication: Treat agentic AI as a distinct governance class and align controls to runtime identity, not to human access review patterns.

Breaches seen in the wild

• Millions of Misconfigured Git Servers Leaking Secrets: Nearly 5 million misconfigured Git servers expose sensitive secrets and credentials online.
• Massive Docker Hub Secrets Leak: 10,000+ Docker Hub container images expose hardcoded secrets and authentication keys.

Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Agentic AI identity sprawl is now an IAM governance problem, not just a tooling gap. The article shows that AI agents are generating dozens of non-human identities and accessing multiple enterprise systems through credentials that existing IAM was never built to manage. That makes identity sprawl dynamic rather than static, with the governance burden shifting from account administration to lifecycle orchestration. Practitioners should stop treating these identities as incidental integrations and classify them as governed subjects in their own right.

Ephemeral-looking agent access still behaves like persistent privilege if it is built on reusable credentials. The article's hardcoded-credential warning is really about privilege persistence disguised as automation. When an AI agent can repeatedly reuse the same secret across workflows, the organisation has not reduced standing access, it has merely hidden it inside software. The practical implication is that credential lifetime becomes a core control variable for agentic AI.

Access review processes assume a subject exists long enough to be reviewed; agentic AI can invalidate that assumption. That assumption was designed for human and stable machine accounts with observable entitlement states. It fails when the actor can create, consume, and discard identities continuously during execution. The implication is that governance must move closer to issuance time, because periodic review alone cannot govern identities that appear and disappear inside active workflows.

Time-bound and auditable access is the right policy shape for AI agents because it preserves accountability at machine speed. Omdia's framing, as summarised by 1Password, points toward a control model where access is granted for a task, recorded in a traceable way, and revoked without human delay. That is less about convenience and more about making non-human identity behaviour intelligible to IAM, audit, and incident response teams. Practitioners should treat agentic AI as a lifecycle governance discipline, not a feature toggle.

Named concept: identity sprawl with runtime creation pressure. This article describes a specific pattern where one agentic workflow can spawn multiple identities across applications, keys, and delegated access paths. The result is not just more identities, but more live governance objects than manual IAM processes can track. Security teams need to measure that growth as a programme risk, because the control failure is structural rather than incidental.

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 Guide

What this signals

Identity sprawl will widen before it stabilises. As agentic AI moves from experiments into production workflows, the number of identities to inventory, govern, and revoke will keep rising faster than manual IAM processes can absorb. Security teams should expect the pressure point to shift from initial authentication to lifecycle ownership and revocation discipline.

Runtime creation pressure: When one workflow can spawn multiple non-human identities on demand, governance has to measure how many identities are created per task and who owns each one after execution ends. That is the metric that determines whether access control is still meaningful or merely documented.

53% of security leaders expect AI to run major portions of their infrastructure autonomously within the next three years, according to the 2026 Infrastructure Identity Survey. That shifts the programme question from whether AI agents belong in IAM to whether current control models can survive at that pace.

For practitioners

• Inventory agentic workflows and their identities Document every AI agent, service account, API key, password, and delegated token used by each workflow. Include where the identity is created, which systems it touches, and who owns its lifecycle.
• Eliminate hardcoded agent credentials Replace plaintext and embedded secrets with centrally managed issuance, scoped access, and revocation logic that can be enforced without developer-side workarounds.
• Bound access to the task, not the agent Grant AI agents only the permissions required for a specific workflow window, then revoke or expire those permissions automatically when the task completes.
• Add audit trails for agent actions Record which agent initiated each access request, what data or application it touched, and which downstream action it triggered so security teams can reconstruct activity without guessing.

Key takeaways

• AI agent adoption is turning identity governance into a sprawl management problem because each workflow can create multiple non-human identities and access paths.
• The article's central warning is that static and hardcoded credentials keep privilege alive inside software long after the task should have ended.
• IAM teams need task-bounded issuance, auditable access, and clear lifecycle ownership if they want agentic AI to remain governable.

Key terms

• Agentic AI: Autonomous AI systems capable of planning, deciding, and taking actions, including calling APIs, writing code, and orchestrating other agents, with minimal human oversight. Agentic AI introduces new NHI risks as agents must authenticate to external services.
• Non-Human Identity (NHI): A digital identity assigned to a non-human entity such as a software application, service account, API key, bot, machine, or AI agent that enables it to authenticate and interact with systems without direct human involvement. NHIs now outnumber human identities in most enterprises by 25 to 50 times.
• 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.
• Task-bounded Access: Task-bounded access is a control pattern where permissions exist only for the duration of a specific job and are removed automatically when the job ends. It is especially relevant for AI agents because autonomous systems can otherwise retain or recreate access long after the original need has passed.

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.