TL;DR: Non-human identities now outnumber humans 45 to 1 in many enterprises, and Token Security argues that AI agents are accelerating both identity creation and overprivileged access, turning machine credentials into a primary attack surface. Static IAM, periodic audits, and human-centric governance no longer match machine-speed identity sprawl.
At a glance
What this is: NHI Mgmt Group's analysis, drawing on Token Security's blog post: Non-human identity security risks are rising in AI enterprises, with rapid NHI growth, overprivilege, and machine-speed identity creation driving the problem.
Why it matters: It matters because IAM, IGA, and PAM programmes that still assume human-paced access management will miss the volume, volatility, and ownership gaps of NHIs and AI agents.
Context
Non-human identity security risk is the governance gap that opens when machine credentials, service accounts, API keys, bots, and AI agents scale faster than ownership, review, and revocation processes. In AI enterprises, the problem is no longer just identity volume. It is that access is created and consumed at machine speed while control functions still assume human-paced administration.
The article frames this as a shift from static IAM to dynamic machine identity governance. That matters because autonomous agents, ephemeral workloads, and CI/CD automation do not fit human lifecycle models, and they often receive broad access to keep workflows moving. The result is an expanding attack surface that is difficult to inventory, certify, and constrain manually.
Key questions
Q: What breaks when non-human identities have more access than they need?
A: When non-human identities carry excess access, a single compromise can move from a local incident to broad cloud control. Over-privileged service accounts, tokens, and AI agents can reach storage, compute, and IAM functions that were never necessary for their job. The result is larger blast radius, faster lateral movement, and much harder containment.
Q: Why do AI agents increase non-human identity risk?
A: AI agents increase non-human identity risk because they can execute many actions quickly once they inherit a credential or tool permission. That speed expands blast radius, shortens attacker dwell time, and makes weak delegation more dangerous. The remedy is tighter scoping, continuous verification, and strict separation between observation and execution privileges.
Q: What are the signs that machine identity management is failing in an organisation?
A: Common signs include incomplete inventory, spreadsheet based tracking, manual renewal processes, unclear ownership, and repeated certificate expiry events. Another signal is when teams struggle to audit where machine identities exist or cannot automate lifecycle actions at scale. If operational teams keep reacting to expiring certificates instead of governing identity lifecycles proactively, the control environment is already under strain.
Q: How can organisations reduce the blast radius of compromised build credentials?
A: Use short-lived credentials, separate build and deploy privileges, and require audited approval paths for production changes. If a build identity is compromised, it should not be able to move laterally into secrets stores or cloud administration. The goal is to make one exposed token useful for only one narrow task.
Technical breakdown
Why machine identity volume breaks human IAM assumptions
Traditional IAM was built around people, not software entities. Humans have stable lifecycles, visible ownership, and interactive authentication flows such as passwords and MFA. NHIs behave differently: they multiply with microservices, cloud-native workloads, and automation, and they often authenticate non-interactively with tokens, certificates, or API keys. That creates a governance problem, because the environment can contain thousands of machine identities whose access is valid, active, and poorly attributed even when no human remembers creating them. In practice, the control plane becomes the identity inventory itself.
Practical implication: Treat NHI discovery and ownership mapping as a core identity control, not an optional audit exercise.
How agentic AI accelerates NHI sprawl
AI agents change the rate and shape of identity creation. Instead of a developer provisioning a small number of accounts, an agent can instantiate infrastructure, create roles, request credentials, and chain sub-agents within seconds. That introduces machine-speed expansion into environments where reviews, approvals, and access certification are still designed for slower human workflows. The issue is not only scale, but unpredictability: access needs emerge at runtime, and privilege is often granted broadly to avoid task failure. This is where AI turns NHI management from lifecycle administration into continuous governance.
Practical implication: Design for runtime identity issuance and revocation, not quarterly review of already-stable credentials.
Why overprivilege and secret sprawl are the main failure modes
Overprivileged service accounts and leaked secrets are the highest-leverage weaknesses in NHI programmes. A credential that exists in code, logs, or a pipeline can be reused anywhere unless it is tightly scoped and rotated. When the same identity also has broad cloud or data access, compromise becomes a blast-radius problem rather than a single-account problem. The article also highlights a common governance failure: temporary access is often granted as permanent privilege, because static roles are easier to manage than task-scoped issuance. That mismatch is what makes machine identities attractive to attackers.
Practical implication: Reduce blast radius by tightening scope, rotating secrets, and removing permanent privilege from temporary workflows.
Threat narrative
Attacker objective: Use valid machine access to reach sensitive data, cloud controls, or software supply chain paths with minimal resistance.
- Entry occurs when attackers find exposed machine credentials in repositories, logs, pipelines, or poorly governed integrations, giving them valid access without needing to phish a person.
- Escalation happens when the compromised NHI already has overprivileged access, allowing the attacker to move from one service or workload to broader cloud and data resources.
- Impact follows when that identity is used to exfiltrate data, tamper with workloads, or propagate across machine-to-machine trust relationships before detection.
Breaches seen in the wild
- Internet Archive breach 2024: An exposed GitLab token opened Internet Archive code and 31 million user records; unrotated Zendesk tokens let the attacker back in weeks later.
- Salesloft OAuth token breach: hackers stole OAuth tokens to access Salesforce data via Salesloft.
Read our 52 NHI Breaches Analysis report for a comprehensive view of breaches impacting Non-Human Identities including AI Agents.
NHI Mgmt Group analysis
Non-human identity sprawl is now a governance problem, not a tooling problem. The article shows that machine identities are not just more numerous than human accounts, they are created, consumed, and forgotten across cloud, CI/CD, and AI workflows. That means ownership, review, and lifecycle controls are failing at the programme level, not just at the implementation layer. Practitioners need to treat NHI inventory quality as a prerequisite for identity governance.
Static access models do not fit machine-speed identity behaviour. Access models built for employees assume a relatively stable subject with a reviewable privilege history. AI agents and ephemeral workloads break that assumption because they can create, use, and discard credentials faster than certification cycles can observe them. The implication is that governance has to move from periodic attestation to continuous issuance control.
Overprivilege is the real multiplier in NHI compromise. The article repeatedly connects machine identity risk to broad roles, unrotated secrets, and unclear ownership. That combination turns one exposed credential into environment-level reach. In practice, blast-radius control becomes the central design objective for NHI governance.
Credential lifecycle is the new perimeter for AI enterprises. The article makes clear that API keys, tokens, and service accounts now sit where firewalls used to sit: at the boundary between safe operation and uncontrolled access. If those credentials are long-lived or hard to attribute, the enterprise has already accepted avoidable risk. Security teams should treat lifecycle enforcement as the control that stands in for perimeter trust.
Machine identity governance has become foundational to AI readiness. Autonomous systems rely on identities to access tools, resources, and data, which means AI risk is inseparable from identity risk. The field should stop framing NHI security as an adjacent discipline and start treating it as core operational governance for AI-enabled enterprises. The practitioner conclusion is simple: no AI programme is safer than its machine identity control plane.
From our research library:
- NHIs outnumber human identities by 25x to 50x in modern enterprises, according to the Ultimate Guide to NHIs.
- 1 in every 1,000 non-human identities in enterprise environments is more than 10 years old.
- Read next: Ultimate Guide to NHIs
What this signals
Identity blast radius: the practical unit of NHI risk is no longer a user account but the combined reach of credentials, roles, and automation paths attached to it. When a single machine identity can reach production systems, data stores, and deployment tooling, governance has to measure the downstream impact of compromise rather than only the presence of access.
Access reviews assume privilege lasts long enough to be observed and certified. AI agents and ephemeral workloads can acquire and release access inside a single task, so the control point shifts from periodic attestation to issuance-time governance.
The main programme risk is not simply more identities. It is unmanaged identity creation without equally automated revocation, ownership, and behavioural monitoring. That is where NHI governance starts to diverge from human IAM in operational terms.
For practitioners
- Map every machine identity to an owner Build an authoritative inventory across cloud, CI/CD, SaaS, and AI workflows, then attach a named owner and business purpose to each credentialed entity.
- Right-size broad roles on service accounts Review machine identities that carry admin or editor permissions but only perform narrow tasks, then remove unused access paths and document the minimum required scope.
- Rotate secrets automatically and shorten credential lifetime Replace static tokens and keys with ephemeral issuance where possible, and set rotation so unused credentials cannot survive long enough to become forgotten attack paths.
- Detect anomalous machine-to-machine behaviour in runtime Monitor for service accounts and agents that suddenly access unfamiliar data stores, new regions, or new workflows, because behavioural drift often precedes credential abuse.
- Separate temporary task access from standing privilege Use just-in-time access for machine tasks that do not require persistent entitlement, especially where AI agents or pipelines need short-lived access to sensitive systems.
Key takeaways
- Non-human identity risk in AI enterprises is rising because machine credentials are being created faster than ownership and lifecycle controls can keep up.
- The article ties the problem to broad access, leaked secrets, and automation, with human-centric IAM no longer matching machine-speed identity behaviour.
- The practical response is to inventory every machine identity, reduce standing privilege, and make rotation and behavioural monitoring continuous.
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 sets the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Non-Human Identity Top 10 | NHI-01 — Improper Offboarding | The article stresses orphaned machine identities that persist after workloads retire. |
| NHI-05 — Overprivileged NHI | Overprivileged service accounts and agents are central to the article's risk model. | |
| NHI-07 — Long-Lived Secrets | The article repeatedly points to static tokens and keys that remain valid too long. | |
| Recommendation — Audit and revoke machine identities when the workload, pipeline, or agent that owns them is retired. Right-size NHI privileges so machine identities only retain the access needed for each task. Replace long-lived machine secrets with short-lived credentials and automated rotation. | ||
| NIST SP 800-53 Rev 5 | IA-5 — Authenticator Management | Credential lifecycle management is central to the article's NHI guidance. |
| Recommendation — Apply authenticator management to rotate, expire, and revoke machine credentials on a defined schedule. | ||
| MITRE ATT&CK | TA0006;TA0008 — Credential Access; Lateral Movement | The article describes exposed credentials enabling movement across cloud and supply chain paths. |
| Recommendation — Map exposed machine credentials to credential access and lateral movement risks in detections and hunting. | ||
Key terms
- 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.
- Machine Identity Sprawl: Machine identity sprawl is the uncontrolled growth of non-human identities across teams, platforms, and business processes. It becomes a governance problem when identities are created faster than they can be inventoried, reviewed, rotated, or retired, leaving security teams with incomplete visibility and weak accountability.
- Privileged non-human identity: A privileged non-human identity is any service account, API key, token, certificate, workload, or AI agent that can reach sensitive systems and perform high-impact actions. The risk comes from the access it carries, not from whether a person is operating it directly. Governance must cover lifecycle, scope, and attribution.
- Credential Lifecycle: Credential lifecycle is the process of issuing, rotating, expiring, and revoking secrets, certificates, and tokens across their usable life. For non-human identities, lifecycle discipline is the core control that separates temporary access from persistent exposure.
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 July 5, 2026.
Updated on October 6, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org