Veza’s identity graph reframes privilege, NHI, and agentic AI

At a glance

What this is: This is a vendor analysis of how identity graphs can unify privileged access, NHI governance, data system access, and AI agent controls into one view of effective permissions.

Why it matters: It matters because IAM teams cannot govern blast radius, inheritance, and entitlement drift across humans, NHIs, and agents if they only see fragments of access.

By the numbers:

• 96% of organisations store secrets outside of secrets managers in vulnerable locations including code, config files, and CI/CD tools.
• Only 5.7% of organisations have full visibility into their service accounts.
• NHIs outnumber human identities by 25x to 50x in modern enterprises.

👉 Read Veza's analysis of privileged access, NHI governance, and AI agents

Context

Identity governance breaks down when access is spread across cloud, SaaS, on-prem, and machine workflows faster than teams can reconstruct who can do what. In that environment, effective permissions matter more than assigned roles, because inherited, federated, and dormant access often determine the real blast radius.

The article's primary identity theme is not product functionality but governance visibility across privileged access, non-human identities, data systems, and AI agents. For IAM programmes, the key question is whether you can continuously prove entitlement truth across those actor types rather than infer it from disconnected tools.

Key questions

Q: How should security teams govern effective permissions instead of just assigned roles?

A: Security teams should base reviews on effective permissions, meaning the access an identity can actually exercise after inheritance, federation, and role assumption are applied. That requires tracing reachable resources, not just looking at the original entitlement record. When this view is missing, certification can approve access that is already broader than intended.

Q: Why do non-human identities increase identity blast radius?

A: Non-human identities often run continuously, hold broad permissions, and connect multiple services at machine speed. If one is compromised or left overprivileged, it can open access to many downstream systems before anyone notices. That makes blast radius a better risk metric than raw asset count.

Q: What breaks when AI agents are monitored like ordinary automated jobs?

A: What breaks is the assumption that fixed schedules and static logs are enough. Autonomous agents can act continuously, switch context, and combine tools in ways that simple job monitoring will miss. Security teams need observability that links identity, action, decision context, and oversight state in a single reviewable record.

Q: Why do service accounts and AI agents need different controls from human users?

A: Service accounts and AI agents authenticate and act without the predictable patterns that human identity systems expect. They can operate across runtimes, scale quickly, and carry permissions into automated workflows. That means access decisions should consider workload context, runtime behaviour, and time-bound authority rather than relying only on user-centric IAM patterns.

Technical breakdown

Why effective permissions matter more than assigned roles

Assigned roles describe intent at provisioning time, but effective permissions show what an identity can actually do after inheritance, federation, cross-account assumption, and local overrides are applied. In complex estates, the difference is material: a user, service account, or workload may look constrained on paper while retaining broad operational reach in practice. That is why access graphs are increasingly used to model real blast radius rather than static policy. For IAM and NHI programmes, the technical issue is not only what was granted, but what is still reachable through chained trust and role assumption.

Practical implication: review effective permissions, not just assigned entitlements, before certifying access or shrinking blast radius.

How non-human identities accumulate hidden privilege

NHIs often gain privilege through service account sprawl, embedded credentials, impersonation rights, and workload-to-cloud role assumption. The risk compounds because these identities are frequently unmanaged, lack clear ownership, and are not tied to the same lifecycle controls used for human access. Once an NHI can assume roles across environments, the effective access path becomes much broader than the original credential suggests. This is why visibility into owners, expiration, rotation, and inherited rights is foundational to NHI governance, not an optional hardening layer.

Practical implication: map every NHI to an owner, rotation state, and assumed-role path before treating it as governed.

What agentic AI changes about identity control

AI agents are not just another workload because they can call APIs, fetch data, and automate actions inside live business processes. The identity problem shifts from static authorisation to runtime boundary control, where an agent can accumulate privileges, impersonate across accounts, or expose data through delegated access paths. That makes time-bound access and explicit impersonation tracing more important than broad, persistent permissions. For autonomous or semi-autonomous systems, the key architecture question is whether the programme can prove what the agent may do at runtime, not simply what it was intended to do at design time.

Practical implication: classify agent access separately from generic workload access and constrain it with time-bound, traceable authorisation.

Threat narrative

Attacker objective: The attacker seeks to turn fragmented identity visibility into broad operational reach across cloud, SaaS, and data environments.

1. Entry occurs when privileged access is granted through over-permitted cloud roles, forgotten admin accounts, inherited SaaS rights, or exposed NHI credentials.
2. Escalation follows when those identities can assume additional roles, inherit hidden permissions, or move laterally through SaaS, cloud, and data systems.
3. Impact is achieved by expanding blast radius across sensitive systems, making audit failures, data exposure, and privilege abuse harder to contain.

Breaches seen in the wild

• Meta AI Instagram Account Takeover — 20,225 Instagram accounts hijacked via compromised Meta AI support chatbot with overprivileged access.
• Replit AI Tool Database Deletion — Replit vibe coding AI assistant deletes live production database and creates 4,000 fake user records.

Read our 52 NHI Breaches Analysis report for a comprehensive view of breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Effective-permissions truth is now the governing layer, not a reporting enhancement. The article reflects a shift that many identity programmes still underweight: assigned access is only the starting point, while inherited and assumed access define the real control surface. That matters across privileged users, NHIs, and AI agents because all three can hold more power in execution than they appear to hold in policy. Practitioners should treat entitlement truth as the basis for governance, not the output of it.

Non-human identity sprawl is no longer separable from privileged access governance. Service accounts, tokens, and workload identities now participate in the same privilege lattice as human admins, which means PAM and NHI controls increasingly converge around the same blast-radius problem. If teams govern NHIs in a different operational silo, they will miss the shared inheritance paths that make compromise scalable. The practical conclusion is that NHI governance must be measured in the same effective-access terms used for privileged users.

Agentic AI introduces runtime privilege decisions that static identity models were not built to absorb. AI agents do not merely consume access, they can chain actions, call tools, and expand scope during execution, which changes the meaning of least privilege. The familiar assumption that access can be fully described at provisioning time no longer holds once the actor can choose actions dynamically. Security teams should treat agent runtime behaviour as a first-class identity variable, not a downstream automation concern.

Blast radius has become the most defensible unit of identity security measurement. When access spans SaaS, cloud, data, and delegated machine identities, the question is not whether an account exists but how far a compromise can travel. The article is strongest when it shows that visibility and governance must be built around reachable permissions, ownership, and time-bounded access. Organisations that cannot measure blast radius cannot credibly claim least privilege.

From our research:

• Only 5.7% of organisations have full visibility into their service accounts, according to the Ultimate Guide to NHIs.
• 91.6% of secrets remain valid five days after the targeted organisation is notified, according to the Ultimate Guide to NHIs.
• For lifecycle depth, the NHI Lifecycle Management Guide shows how provisioning, rotation, and offboarding change when machine identities outnumber people.

What this signals

Blast-radius governance is becoming the practical centre of identity security. As cloud, SaaS, and machine identities converge, teams need a model that measures what an account can reach, not just what it was granted. That shifts prioritisation toward effective permissions, ownership, and role assumption paths, especially for NHIs and privileged users.

With 96% of organisations storing secrets outside secrets managers in vulnerable locations such as code, config files, and CI/CD tools, per the Ultimate Guide to NHIs, visibility work has to start before policy tuning. The programme signal is clear: if you cannot see where credentials live, you cannot govern the identities that use them.

Runtime identity control will matter more as AI agents enter operational workflows. The next governance gap is not simply more access, but access that changes during execution. Teams that extend entitlement graphs to agents, workloads, and delegated access paths will be better positioned to contain scope creep before it becomes incident response.

For practitioners

• Map effective permissions across the identity graph Rebuild access reviews around what identities can actually reach after role assumption, inheritance, and federation are applied. Use the result to identify hidden admin paths, dormant accounts, and over-permitted NHIs before certification cycles begin.
• Assign owners to every non-human identity Require a named owner, expiration state, and rotation expectation for each service account, token, and workload identity. Treat unowned NHIs as governance defects, not just hygiene issues, because they cannot be offboarded or attested reliably.
• Separate agent access from generic workload access Classify AI agents by runtime decision scope, then constrain their access with time-bound permissions and explicit impersonation tracing. This prevents agents from inheriting broad standing rights that were designed for static automation.
• Use blast-radius reviews for privileged and cross-cloud access Prioritise identities that can assume AWS, GCP, Azure, or SaaS admin rights and trace the permissions they inherit in each environment. Pair those reviews with remediation targets that reduce the reachable set, not just the number of accounts.

Key takeaways

• The core problem is not identity volume alone, but the gap between assigned access and effective permissions.
• The strongest evidence in this space still points to poor visibility, over-permissioned NHIs, and hidden inherited access paths.
• Practitioners should rebuild governance around blast radius, ownership, and runtime control across people, machines, and agents.

Key terms

• Effective Permissions: Effective permissions are the access an identity can actually use after role inheritance, scope, and policy are applied. In Azure AI environments, they often matter more than the assigned role name because inherited rights can widen access to data, logs, and secret stores.
• Blast Radius: The potential scope of damage if a specific credential or identity is compromised. Identities with broad permissions have a larger blast radius and represent a higher priority for least-privilege enforcement and security controls.
• Identity Graph: An identity graph is a relationship map that connects identities, assets, data, and permissions so teams can see how access actually flows. In NHI programmes, it helps explain which agent is related to which owner, which system, and which policy boundary.
• Agentic AI Identity: The complete set of credentials, permissions, and governance controls applied to an autonomous AI agent — covering authentication, authorisation, action logging, and access revocation. Distinct from traditional NHI because agent identities are often ephemeral, delegated, and multi-hop.

What's in the full article

Veza's full article covers the operational detail this post intentionally leaves for the source:

• How the Access Graph is used to calculate effective permissions across SaaS, cloud, data, and on-prem systems
• Examples of privileged access monitoring workflows for admins, contractors, guest accounts, and federated identities
• How NHI discovery, owner assignment, credential rotation, and vaulting are implemented in the vendor workflow
• The agentic AI use case and plain-language queries for understanding what agents can do and revoke

👉 The full Veza article covers access graphs, effective permissions, and the operational view of blast radius.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity security are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are responsible for identity security strategy or NHI governance in your organisation, it is worth exploring.