Omada Agent Governance shows the identity gap for AI agents

At a glance

What this is: Omada Identity's announcement frames AI agent governance as an identity problem, arguing that existing tools leave gaps in visibility, ownership, access scope, and audit evidence for non-human actors.

Why it matters: It matters because IAM, IGA, and PAM teams need governance models that can cover AI agents alongside service identities and human users without losing accountability or control evidence.

Context

AI agent governance is the practice of assigning ownership, access scope, and review evidence to software entities that act with some operational autonomy. In this announcement, Omada Identity argues that current identity programmes still treat people as the default subject, even as AI agents begin connecting to systems, accessing data, and executing work.

The governance gap is not simply that AI agents are hard to inventory. It is that the surrounding IAM and IGA model was built around stable human identities, while agent behaviour can change faster than review cycles and can span multiple systems at once. That creates a control problem for accountability, access certification, and audit readiness.

Key questions

Q: What breaks when AI agents are not governed at runtime?

A: Without runtime governance, an agent can shift behaviour after provisioning and still execute actions that were never reviewed in context. That is where tool chaining, MCP connections, and rapid decision-making become dangerous. Static approval cannot stop a live change in intent, so teams lose control at the point of action.

Q: Why do AI agents create new risk in non-human identity management?

A: AI agents create risk because they operate as software identities with delegated authority, but many organisations do not track them with the same discipline applied to users or service accounts. They can connect quickly, persist across teams, and accumulate permissions that are hard to review. That combination increases the chance of unnoticed access drift and credential exposure.

Q: How do security teams know if an AI agent has too much access?

A: Look for agents that can reach multiple systems without task-specific limits, use persistent tokens, or touch high-value services such as email, chat, cloud consoles, and file stores. A healthy deployment leaves a clear audit trail of what the agent can do, what it actually did, and which credentials it used.

Q: What should organisations do before scaling AI agent access?

A: Organisations should define ownership, consent, auditability, and revocation for agents before broad deployment. If the agent can access customer accounts, the control set must be in place at design time, not after the first incident. Scaling without those controls turns experimentation into governance debt.

How it works in practice

Why AI agent governance breaks human-centred identity models

Human-centric identity controls assume a known subject, a defined role, and a relatively stable access profile. AI agents disrupt that pattern because the actor can be instantiated, copied, delegated, or retired far more quickly than an employee account lifecycle. The practical difficulty is not only discovery, but mapping ownership and acceptable access at the moment the agent is created. Once an agent can execute tasks across systems, the governance question shifts from who logged in to who is responsible for a machine actor's reach and decisions.

Practical implication: model AI agents as governed identities from creation through retirement, not as ad hoc application features.

How access scope and actual usage diverge for non-human identities

The article points to a key control weakness: access granted to AI agents may not match actual usage. That is a classic governance problem for non-human identities, because entitlement assignment often expands faster than consumption patterns are reviewed. Over-privilege becomes harder to spot when an agent can invoke tools, APIs, and data sources dynamically. In that environment, governance needs to compare entitlement with behaviour, not merely record that access exists.

Practical implication: review AI agent entitlements against observed task execution and remove access that is never used.

Why audit evidence becomes harder when agents act across systems

Audit evidence for AI agents has to show ownership, access decisions, and dependencies in a way that is consistent enough to withstand review. That is harder than for people because agent activity may involve multiple platforms, delegated permissions, and ephemeral execution paths. Without clear evidence of who approved what and why, governance turns into retrospective explanation instead of control. The article's emphasis on reliable evidence signals that audit readiness is becoming a design requirement, not a reporting afterthought.

Practical implication: retain evidence that ties each AI agent to an owner, an access decision, and a review record across its lifecycle.

NHI Mgmt Group analysis

AI agent governance is becoming a distinct identity discipline, not a feature of existing IGA. The article is right to frame AI agents as a new class of digital actor rather than a minor extension of workload identity. Once an actor can connect, access, execute, and make decisions, ownership and entitlement management need their own lifecycle treatment. Practitioners should stop treating agent oversight as an add-on to user governance and start treating it as a governed identity class.

Identity governance fails when accountability is not first-class for non-human actors. The article's four questions, what agents exist, who owns them, what they can reach, and what risk they introduce, describe the real governance boundary. Without explicit accountability, orphaned agents become the machine equivalent of unowned privileged accounts. That means the governance problem is not visibility in isolation, but the absence of durable ownership attached to non-human execution.

Over-privilege is the central control issue because AI agents can outpace review cycles. The article points to the gap between granted access and appropriate access, which is where governance failures usually begin. AI agents can be spun up quickly and used across multiple systems before certification processes catch up. Practitioners should treat dynamic execution as a reason to narrow standing access, not as evidence that broader access is justified.

Audit readiness now depends on evidence that links identity, action, and approval. Omada's focus on reliable governance evidence reflects a broader market shift: auditors will increasingly expect a traceable chain from agent ownership to access decision to observed activity. That requirement cuts across NHI governance, IGA, and AI governance. The practical conclusion is that evidence quality, not just policy presence, will determine whether AI agent governance holds up under scrutiny.

AI agent governance should be anchored in a named control concept: governance evidence continuity. The article makes clear that organisations need a continuous record of who is accountable, what access exists, and how it is used. That continuity is what turns identity governance from a point-in-time approval exercise into a defensible control system. For practitioners, the lesson is to design for uninterrupted evidence across the full AI agent lifecycle.

From our research library:

• 19% of organisations give AI systems dramatically more access than human employees, nearly one in five granting unrestricted privilege, according to the 2026 Infrastructure Identity Survey.
• Systems with least-privileged AI access had a 17% incident rate vs 76% for over-privileged systems. Organisations failing to scope AI access properly are 4.5x more likely to experience a security incident, according to the 2026 Infrastructure Identity Survey.
• Read next: NHI Lifecycle Management Guide

What this signals

Governance evidence continuity: AI agent programmes need a continuous record that ties ownership, access decisions, and observed behaviour together. Without that chain, review activity becomes a paperwork exercise that cannot prove whether an agent should still be operating. For IAM and IGA teams, the control question shifts from approval alone to whether evidence survives the full lifecycle.

AI agents change the unit of governance from account to actor. That matters because entitlement models built for people assume a stable identity history, while agent identities can be created, delegated, and retired much faster. Practitioners should prepare for governance workflows that follow the actor across systems instead of only tracking login events.

For practitioners

• Define AI agents as governed identities Create a governance category for AI agents with explicit ownership, lifecycle status, and accountable approvers so they are not tracked as generic application components.
• Map agent ownership and accountability Require a named business and technical owner for every agent before it is allowed to access systems or data, and block orphaned agents from production use.
• Compare granted access with actual usage Review whether each agent's entitlement set matches the tasks it actually performs, then remove access that is never exercised or no longer needed.
• Retain audit evidence across the agent lifecycle Preserve access approvals, ownership records, dependency mappings, and usage evidence so reviews can reconstruct why each agent was allowed to operate.
• Align AI agent governance to applicable control frameworks Use policy, risk, and assurance mapping to show how agent governance connects to AI risk management, access governance, and audit readiness requirements.

Key takeaways

• AI agents create a governance gap because current identity programmes still assume the subject is a person with stable ownership and predictable access.
• The article frames the problem as an accountability and evidence issue, not just a visibility issue, which is where IGA gaps usually become operational.
• Practitioners should govern AI agents as lifecycle-managed identities, with ownership, usage checks, and audit evidence attached from creation to retirement.

Key terms

• AI Agent Governance: AI Agent Governance is the set of policies, controls, and oversight practices used to direct how autonomous software agents behave. It defines allowed actions, approval paths, identity boundaries, logging, monitoring, and accountability so agent decisions remain traceable, constrained, and aligned with business, security, legal, and ethical requirements.
• Governance Evidence: The records that prove a control existed and operated when needed. For AI programmes, that usually means logs, approvals, review outcomes, and lifecycle artefacts that show who owned the system, what it accessed, and how it was retired.
• Orphaned Agent: An orphaned agent is an AI or software identity that still has access but no longer has a clear human owner responsible for its governance. The risk is lifecycle drift, where the access remains active after role changes or departures, leaving accountability and revocation processes incomplete.
• Access-Usage Mismatch: A condition where an identity is entitled to more access than its actual work requires or uses. For AI agents, this gap is especially dangerous because dynamic behaviour can hide excessive privilege until governance checks compare granted access with observed execution.

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.