Governing NHI and AI identities inside business applications

At a glance

What this is: This is an analysis of why business application governance built for human users fails when NHIs and AI agents operate continuously and accumulate access across systems.

Why it matters: It matters because IAM, IGA, and PAM teams need governance that covers human, machine, and AI-driven identities before privilege creep and audit gaps become operational defaults.

By the numbers:

• Today, NHIs outnumber human users 45:1 in large enterprises, and the gap is widening.

Context

Traditional identity governance was built around human life events such as joiners, movers, and leavers. That model assumes access changes happen at a pace people can review, approve, and certify.

Saviynt argues that assumption no longer holds when non-human identities and AI agents operate continuously across business applications, cloud, and SaaS. The problem is not just volume, but the mismatch between machine-speed access and human-paced governance.

In this environment, the relevant governance question is not whether identities can be inventoried. It is whether ownership, visibility, and access review still work when the identity subject is not a person.

Key questions

Q: What breaks when non-human identities are managed with human joiner-mover-leaver processes?

A: Human lifecycle models assume a person has a start date, role changes, and an offboarding event. Non-human identities do not follow that pattern, so service accounts and secrets can persist after the business purpose changes. The result is unmanaged standing access, unclear ownership, and delayed revocation across infrastructure.

Q: Why do unstructured non-human identities create more risk than managed applications?

A: Unstructured identities are harder to discover, monitor, and govern than managed applications. They often lack clear ownership, consistent lifecycle controls, and strong guardrails such as MFA. When service accounts, API keys, or SSH keys proliferate across SaaS and cloud platforms, attackers gain more opportunities to abuse standing access and move laterally through trusted integrations.

Q: How do teams know if NHI governance is actually working?

A: Look for complete inventory coverage, clear ownership, enforced rotation, and reliable decommissioning. If new credentials appear faster than they are classified, or if stale secrets stay valid after workload changes, the programme is not governing machine identities effectively.

Q: How should IAM teams govern AI-assisted identity workflows?

A: Treat AI-assisted identity workflows as governed control paths, not simple productivity tools. Define which tasks the assistant may recommend, draft, or execute, then keep approval rights and exception handling with named humans. The essential control is traceability, so every machine-generated action can be reviewed, challenged, and linked back to a responsible operator.

Technical breakdown

Why human-centric joiner-mover-leaver models fail for NHI governance

Joiner-mover-leaver governance assumes a person has stable lifecycle events, a clear manager, and reviewable access changes. Non-human identities do not follow that pattern. Service accounts, bots, and AI agents can be created, modified, delegated, or retired by business processes rather than HR events, which means certification windows often open after the access change has already happened. When the governed subject can act continuously, the control point shifts from periodic review to lifecycle visibility and ownership at issuance time.

Practical implication: move governance signals from calendar-based review cycles to identity creation, delegation, and retirement events.

How cross-application access compounds business application risk

Business applications such as SAP, Oracle, Salesforce, ServiceNow, and Workday do not hold identity risk in isolation. An NHI may inherit permissions in one application, receive additional access in another, and end up with a privilege combination that no single team fully sees. That cross-application accumulation is what turns normal overprovisioning into SoD exposure, because the effective power of the identity is created by relationships between systems, not by any one entitlement alone. The result is an access estate that looks fragmented until it is analysed as one governance surface.

Practical implication: evaluate entitlement combinations across business applications, not just access inside each application.

What continuous monitoring changes when AI agents act at machine speed

Continuous monitoring is not simply a more frequent audit. It is a different control model for identities that can spawn related accounts, delegate work, and keep operating outside human review windows. That matters because quarterly or annual recertification assumes access remains stable long enough to be sampled. For AI agents and other NHIs, that assumption breaks. Governance has to detect policy violations, anomalous behaviour, and overprovisioning while the identity is still active, otherwise the control only documents exposure after the fact.

Practical implication: design detection and remediation around live identity behaviour rather than post-cycle certification.

Threat narrative

Attacker objective: The objective is to preserve unchecked access paths across business applications long enough to exploit hidden privilege combinations, policy violations, or unreviewed automation paths.

1. Entry occurs when an NHI or AI agent is introduced into business applications without full inventory, ownership, or lifecycle oversight.
2. Credential or privilege access expands as the identity inherits permissions or is granted access broader than the task requires.
3. Escalation happens when access accumulates across applications and the identity reaches a risky combination that creates SoD violations or hidden reach into corporate resources.
4. Impact is prolonged exposure, because excessive access and policy violations remain active long after the business task is complete.

Breaches seen in the wild

• Coupang Signing Key Breach: Unrevoked signing key credentials expose 33.7 million records after employee offboarding failure at Coupang.
• Ledger Connect Kit npm compromise 2023: Attackers phished a former Ledger employee with unrevoked npm access and published a wallet-draining Connect Kit, stealing about $600k on 14 Dec 2023.

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

Human-centric IGA is now a partial control plane, not a complete one. It still has value for joiner-mover-leaver governance, but only for identities that behave like people. The moment the governed subject is a service account, bot, or AI agent, the operating assumptions change: no HR lifecycle, no stable manager, and no useful quarterly review cadence. Practitioners should stop treating human IGA as the default model for all identities.

Cross-application privilege is the real exposure surface in business applications. The article is right to focus on SAP, Oracle, Salesforce, ServiceNow, and Workday because those environments turn isolated permissions into compound access. A single entitlement may be harmless, but the combination across applications can create SoD violations and hidden privilege paths. The governance problem is therefore relationship mapping, not just entitlement counting.

Continuous governance is becoming the baseline for machine-speed identities. If an identity can be created, delegated, and retired within minutes or hours, any control that depends on quarterly sampling is structurally behind. That is true for NHI lifecycle oversight and becomes even more pronounced when AI agents can trigger new identities as part of their work. Practitioners need continuous visibility, not periodic reassurance.

Shadow AI turns identity discovery into a governance control, not an inventory exercise. The article highlights that some NHIs are known while others emerge organically inside business applications and automation flows. That means discovery is no longer just an asset-management task. It is the prerequisite for ownership, certification, and policy enforcement, especially when no business owner has been assigned.

Consistent governance across human, non-human, and AI identities is the only durable model. Separate policy stacks for each identity type will not close the blind spots created by mixed execution environments. The stronger pattern is identity-agnostic governance with different operating rules for different actor types. Practitioners should design one governance architecture that can classify and control all three identity classes without assuming they behave alike.

From our research library:

• NHIs outnumber human identities by 25x to 50x in modern enterprises, according to the Ultimate Guide to NHIs.
• Read next: Ultimate Guide to NHIs

What this signals

Identity-agnostic governance is becoming the practical baseline: business application teams can no longer assume that review cadences built for humans will control identities that act continuously. The programme shift is from periodic approval to continuous visibility over issuance, delegation, and retirement.

Shadow AI changes the security operating model: once AI capability can appear inside a workflow without a clear owner, discovery becomes a prerequisite to governance. Practitioners should treat unowned automation as an access-risk problem, not just an application-management issue.

For practitioners

• Map every non-human identity to a named business owner Require ownership for service accounts, bots, and AI agents before they are allowed to touch production applications. If no accountable owner exists, the identity should not pass governance review.
• Replace periodic reviews with event-based governance triggers Trigger recertification when an NHI is created, delegated, expanded, or retired, rather than waiting for quarterly or annual review cycles that miss machine-speed change.
• Assess cross-application privilege combinations Review the combined access footprint across SAP, Oracle, Salesforce, ServiceNow, Workday, cloud services, and SaaS so that hidden SoD conflicts do not remain invisible inside separate application teams.
• Inventory shadow AI and unowned automation paths Find AI capabilities embedded inside business applications and workflows, then classify whether they are known, owned, and governed or operating outside formal identity oversight.
• Apply one governance model across all identity types Use a common control architecture for human users, non-human identities, and AI agents, while still preserving actor-specific policy rules for review cadence, lifecycle events, and remediation.

Key takeaways

• Human-centric joiner-mover-leaver controls do not translate cleanly to service accounts, bots, and AI agents that do not follow employee lifecycle events.
• The strongest risk is cross-application privilege accumulation, where access becomes dangerous only when entitlements are combined across systems.
• IAM and IGA teams need continuous visibility, named ownership, and event-based control points if they want governance to keep pace with machine-speed identities.

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.
• Cross-application privilege: Cross-application privilege is the effective access an identity gains from the combined permissions it holds across multiple systems. It is often more dangerous than any single entitlement because risk emerges from how permissions interact across business applications, cloud services, and SaaS platforms.
• Shadow AI: AI agents, copilots, or connected tools operating without full visibility or governance from security teams. Shadow AI becomes an identity problem when those systems authenticate with unmanaged tokens, service accounts, or OAuth apps that can reach production resources.
• Segregation of Duties: Segregation of Duties is a control principle that prevents one person or role from combining incompatible permissions that could create fraud, error, or undetected change. In ERP environments, it must account for roles, transactions, approvals, and compensating controls across business processes.

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.