Yes. AI governance becomes much harder when the human provisioning layer is already overloaded with excessive access and unclear review workflows. Organisations should stabilise workforce identity first so agent access is inherited from a governance model that people can actually use and understand.
Why the sequence matters
AI governance works best when the underlying workforce identity model is already clean enough to support clear ownership, review, and escalation. If joiner-mover-leaver workflows are noisy, role definitions are inconsistent, or access reviews are already overloaded, the organisation will simply add another governance layer on top of unresolved control debt. That makes AI oversight slower, less credible, and harder to operate at scale.
The practical reason to sequence this way is that AI governance usually depends on the same approval paths, entitlement records, and reviewer judgement that workforce identity uses. If those records are inaccurate, every downstream decision about who can approve, supervise, or override agent behaviour inherits that weakness. Stabilising the human control plane first gives the AI programme a governance baseline that is understandable and repeatable.
What a stable workforce identity layer gives AI governance
A stable workforce identity layer gives the organisation a reliable picture of who has standing access, who approves it, and when reviews happen. That matters because AI governance often needs human sign-off for sensitive actions, policy exceptions, escalation, and exception handling. If the human process is already fragmented, the AI process will usually be interpreted differently by different teams.
It also reduces ambiguity around authority. A governance model for identity and access management and identity governance should already tell you how access is requested, granted, reviewed, and revoked before you decide how an agent fits into that chain. That is especially important when the organisation expects AI tools to inherit human permissions, because the human role model becomes the reference point for what the agent can do.
For many organisations, the right sequence is not “pause AI until identity is perfect,” but “make identity operationally coherent enough that AI governance does not become an exception factory.” A workforce identity security guide is most useful when the business is already trying to tighten provisioning, authentication, and recovery paths that would otherwise leak into AI oversight decisions.
How to tell when you are ready to move AI governance forward
Readiness is less about policy volume and more about whether the organisation can answer basic questions without debate. Can it name the access owner, explain the review path, and revoke access on time? If those answers are inconsistent for people, they will be inconsistent for agent-adjacent access too. The sequencing decision should therefore be driven by operational clarity, not by enthusiasm for the newest platform.
That is why an identity security programme guide is relevant here: AI governance needs a programme structure, not just a set of controls. Where the organisation is still arguing over ownership, recertification cadence, or who can approve access exceptions, the AI layer will amplify those disagreements rather than resolve them.
A useful practical test is whether the workforce access process can support a review without special casing. If managers, app owners, and security reviewers already rely on ad hoc spreadsheets or informal approvals, they are not ready to govern AI access with confidence. In that case, the right first move is to reduce review ambiguity and entitlement sprawl before introducing new AI-specific governance obligations.
Risk and Threat Considerations
When AI governance is layered on top of an overloaded workforce identity process, the organisation risks normalising weak approval discipline and unclear accountability. That creates a bigger exposure than simple admin friction, because the same reviewers and approvers may end up blessing both human access and agent access without a dependable standard.
Failure mechanism: Excessive standing access, unclear ownership, and inconsistent recertification create a control environment where agent permissions inherit the weakest human process rather than a deliberate governance model.
Impact: The result is higher privilege drift, slower revocation, weaker auditability, and a greater chance that an AI system is trusted to act under permissions the organisation cannot justify or explain.
That concern is closely related to regulatory and audit perspectives on non-human identities, because once governance is inconsistent, the organisation often struggles to evidence why a given access path was allowed in the first place. The more exceptions are tolerated in workforce identity, the easier it is for AI governance to become a paper exercise.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Agentic AI Top 10 addresses the attack surface, NIST SP 800-53 Rev 5 and NIST AI RMF set the technical controls, and ISO/IEC 42001:2023 defines the regulatory obligations.
| Framework | Control / Reference | Relevance |
|---|---|---|
| NIST SP 800-53 Rev 5 | IA-5 — Authenticator Management | Covers the credential lifecycle discipline needed before AI access builds on it. |
| AC-2 — Account Management | Directly applies to joiner-mover-leaver hygiene and access ownership that AI governance inherits. | |
| Recommendation — Tighten credential issuance, rotation, and revocation before extending access paths to AI use cases. Stabilise account provisioning, review, and deprovisioning so AI access inherits a reliable identity baseline. | ||
| NIST AI RMF | GOVERN — Govern | AI governance is the core subject, especially accountability and oversight sequencing. |
| Recommendation — Establish clear AI oversight ownership and policy before expanding agent-enabled access. | ||
| ISO/IEC 42001:2023 | 4.4 — AI management system | Applies because the question is about sequencing AI governance within an organisational control system. |
| Recommendation — Embed AI governance into an auditable management system after foundational identity controls are stable. | ||
| OWASP Agentic AI Top 10 | ASI03 — Identity & Privilege Abuse | Agent access inherits human privilege patterns, making identity and privilege abuse a key AI risk. |
| Recommendation — Constrain agent permissions to prevent inherited privilege abuse from weak workforce identity controls. | ||
Practitioner Guidance
What to prioritise: Clean up the highest-friction parts of workforce identity first, especially access review quality, role clarity, and offboarding reliability. Those are the controls most likely to be reused, directly or indirectly, when AI governance starts assigning ownership and approval responsibility.
Decision rule: If your identity team cannot complete routine human access governance without heavy manual intervention, treat AI governance as a second-phase programme. If the human control plane is predictable, you can extend it to AI with far less ambiguity.
What to verify: Verify that access owners, reviewers, and revocation paths are actually working in practice, not just documented. The question is whether the organisation can sustain the governance load once AI introduces more exceptions, more tooling, and more cross-functional review points.
Practitioner takeaway: AI governance should inherit discipline, not compensate for chaos; if workforce identity is still unstable, the AI programme will spend its time absorbing unresolved access problems instead of governing AI risk.