TL;DR: Akeyless says Gartner published two April notes that together move secrets management toward workload identity, with guidance to avoid static credentials, govern across multiple vaults, and treat AI agents and other workloads as identities rather than secret holders. The shift is from stored credentials to runtime-issued access and central policy.
At a glance
What this is: This is an analysis of how two Gartner notes recast secrets management around workload identity, multi-vault governance, and runtime access for AI agents and other workloads.
Why it matters: It matters because IAM, PAM, and NHI programmes have to govern identities and access at runtime, not just store and rotate secrets after the fact.
👉 Read Akeyless's analysis of Gartner’s workload identity brief and secrets management shift
Context
The core problem is no longer just secret storage. In cloud and AI-heavy environments, static credentials create governance blind spots because workloads and agents need access at runtime, across platforms, and often across multiple vaults.
For IAM and NHI teams, that shifts the control plane from credential custody to identity governance. The article argues that organizations need central policy, inventory, and enforcement across distributed vaults, rather than assuming one repository can absorb the complexity.
The article also treats AI agents as part of the same governance problem when they authenticate and act as workloads. That makes this a workload identity and secrets-management story, not a simple vault consolidation story.
Key questions
Q: What breaks when workloads still rely on static credentials for service-to-service access?
A: Static credentials break down when workloads are ephemeral, distributed across multiple environments, or expected to authenticate without preconfigured secrets. They are difficult to rotate safely, easy to expose in configuration or environment variables, and poorly matched to dynamic infrastructure. The result is brittle access control, weaker auditability, and a larger attack surface for compromise and lateral movement.
Q: Why do multiple secret vaults create governance risk?
A: Multiple vaults create risk when policy, audit, and revocation are fragmented across them. The challenge is not only storage sprawl but inconsistent control enforcement, which makes it harder to know where secrets live and whether the same access rules apply everywhere.
Q: How should security teams govern AI agents that rely on shared runtime credentials?
A: Security teams should treat every AI agent as a workload identity with a defined task boundary, then issue only the minimum access required for that task. Shared credentials should be retired where possible because they obscure accountability, widen blast radius, and make revocation harder after the task completes.
Q: What is the difference between secrets management and workload identity?
A: Secrets management protects stored credentials, while workload identity governs how a workload proves who it is before any secret is issued. Both matter, but workload identity addresses the bootstrap problem that secrets managers cannot solve on their own. In practice, identity should lead and secrets storage should support it.
Technical breakdown
Why static credentials break workload identity governance
Static credentials are durable secrets that outlive the task, session, or workload that uses them. In cloud environments, that persistence creates a mismatch between the identity moment and the access moment: the secret exists long before the workload needs it, and often remains valid after the workload no longer should. Workload identity management changes the model by binding access to the runtime identity of the workload, then issuing short-lived access when policy allows it. That reduces the value of a leaked credential and makes the control point the identity assertion, not the stored secret.
Practical implication: Practitioners should move high-value workloads off durable shared secrets and onto runtime identity-backed access where policy can constrain issuance.
How multi-vault governance differs from vault consolidation
Multi-vault governance assumes organizations will continue running several secret stores across clouds, teams, and platforms. The problem is not merely inventory sprawl, but inconsistent policy, uneven rotation discipline, and fragmented audit trails across those vaults. Orchestration or synchronization can create a governed layer above that sprawl without requiring a disruptive rip-and-replace migration. That matters because central visibility and control are what let security teams understand where secrets live, who can retrieve them, and whether the same policy is being applied everywhere.
Practical implication: Practitioners should prioritize cross-vault policy, audit, and lifecycle controls before attempting a single-vault standardization project.
Why AI agents force a different access model
AI agents can act as workloads, but their access needs are often more dynamic than traditional applications because they can invoke tools, traverse systems, and request credentials during runtime. That means a long-lived API key or token is not just a storage problem but a governance problem: the credential can be reused, copied, or retained beyond the intended task. Workload identity management gives the agent a verifiable identity and short-lived access instead of a persistent secret. The key architectural change is that authorization must be issued for the task, not assumed from a stored credential.
Practical implication: Practitioners should treat AI agents as governed workloads and design for ephemeral access issuance rather than embedded keys or tokens.
Threat narrative
Attacker objective: The objective is to reuse durable workload credentials to gain unauthorized runtime access that outlives the original task or identity context.
- Entry occurs when a workload or AI agent is given a static credential that can be reused outside the intended runtime context.
- Credential exposure then turns that secret into a durable point of misuse, because the secret can be harvested from code, configuration, or vault sprawl.
- Escalation happens when multiple vaults, inconsistent policies, or shared secrets make it hard to constrain where the credential can be used.
- Impact is unauthorized access at runtime, with the added governance failure that revocation and audit become fragmented across tools and teams.
Breaches seen in the wild
- Azure Key Vault Contributor escalation 2024: Datadog found Azure Key Vault Contributor could add itself to access policies and read every secret, key and certificate in a vault.
- Hugging Face Spaces breach 2024: Unauthorised access to Hugging Face Spaces may have exposed secrets users stored for AI apps; tokens were revoked and org tokens removed.
Read and download The State of NHI & AI Agent Breach Report 2026, covering 150+ breaches impacting Non-Human Identities including AI Agents.
NHI Mgmt Group analysis
Workload identity is becoming the governing layer for secrets, not a side topic. The article reflects a broader market shift: secrets management is no longer just about safe storage, but about whether a workload should ever hold a reusable secret in the first place. That changes how teams design identity, audit, and revocation across cloud and AI workloads. The practitioner conclusion is that governance now starts at issuance, not recovery.
Multi-vault governance is the realistic control model for most enterprises. The article’s emphasis on orchestration across multiple secret stores matches what many environments already look like in practice. Centralization by migration often creates more risk than it removes, while unified policy and audit can govern distributed vaults without forcing a rebuild. The conclusion for practitioners is to control the sprawl that exists, not the architecture they wish existed.
Static credential assumptions fail when AI agents and workloads act at runtime. Least privilege was designed for identities whose access scope is reasonably stable across a review period. That assumption weakens when the actor acquires and uses credentials dynamically as part of execution. The implication is not merely more automation, but a redesign of how runtime access is granted, observed, and revoked.
Runtime access is now the decisive control plane for NHI governance. The article points to a category boundary shift from vault administration to workload identity administration. Once identities are assigned to applications, services, and agents, the question becomes whether access can be governed continuously across clouds and vaults. Practitioners should read this as a move toward identity-first control over secrets-first custody.
Workload Identity Management is the right label for a wider governance problem. The named concept matters because it captures inventory, monitoring, administration, and access control in one operating model. That is more useful than treating secrets management, IAM, and AI agent access as separate silos. The practitioner conclusion is to align governance, inventory, and runtime enforcement under a single identity model.
From our research library:
- The agentic AI market is already worth $5 billion and is projected to reach $50 billion by 2030.
- Read next: Agentic AI Identity Guide
What this signals
Runtime identity, not secret custody, is becoming the center of gravity. As more workloads and AI agents need access on demand, the control question shifts from where a credential is stored to whether the identity behind the request is trustworthy at issuance time. Teams that keep treating vault management as the endpoint will miss the governance layer that now matters most.
Multi-vault governance will be the default operating reality. Most enterprises are not moving to one vault, they are moving to distributed vaults with central policy stitched across them. That means the practitioner job is to standardize control, audit, and revocation semantics even when the underlying storage remains heterogeneous.
For practitioners
- Replace durable shared secrets with runtime-issued access Use workload identity and short-lived credentials for applications, containers, and AI agents so a leaked token cannot be reused outside the task it was meant to serve.
- Build governance across existing vaults Map every vault, sync path, and orchestration layer before any consolidation project so policy, audit, and revocation work consistently across cloud and on-premise stores.
- Inventory AI agents as governed workloads Register each agent, the identities it uses, and the systems it can reach so runtime access can be reviewed as part of workload governance.
- Remove embedded API keys from agent workflows Shift agent integrations to ephemeral credential issuance tied to verified workload identity, then block persistent secrets from code, prompts, and environment variables.
Key takeaways
- Static secrets are increasingly a governance mismatch for cloud workloads and AI agents because they outlive the runtime context that should bound them.
- The article’s core evidence is architectural, not operational: the market is moving from secret custody to workload identity and multi-vault control.
- Security teams should redesign access around issuance, inventory, and revocation at runtime rather than assuming vault centralization solves the problem.
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 addresses the attack and risk surface, while NIST SP 800-53 Rev 5 and NIST CSF 2.0 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Non-Human Identity Top 10 | NHI-04 — Insecure Authentication | Static credentials and runtime identity binding are central to the article's governance shift. |
| NHI-07 — Long-Lived Secrets | The article explicitly warns against static credentials and long-lived API keys. | |
| NHI-08 — Environment Isolation | Multi-vault governance and cross-environment access are core to the article's control model. | |
| Recommendation — Replace reusable secrets with identity-bound, short-lived authentication for workloads and agents. Eliminate long-lived secrets from workload paths and issue ephemeral credentials instead. Use environment boundaries and central policy to prevent secrets from crossing unintended trust zones. | ||
| NIST SP 800-53 Rev 5 | IA-9 — Service Authentication | Workloads and AI agents authenticating to services map directly to service authentication controls. |
| Recommendation — Apply IA-9 to authenticate workloads and AI agents with bound, non-shared credentials. | ||
| NIST CSF 2.0 | PR.AA-05 — Access Permissions, Entitlements and Authorizations | The article is fundamentally about governing runtime entitlements across distributed vaults. |
| Recommendation — Define and enforce entitlement policy for workloads at issuance time across every vault. | ||
Key terms
- Workload Identity Management: Workload Identity Management is the practice of creating, issuing, securing, rotating, and revoking identities used by software workloads. It covers how services, containers, functions, and agents prove who they are to other systems, usually through certificates, tokens, keys, or federated assertions, so access can be controlled and audited.
- Multi-vault governance gap: The multi-vault governance gap is the mismatch between a single-vault PAM model and a real enterprise where secrets live in many stores with different rules. It appears when teams can see storage locations but cannot demonstrate control over the full identity path.
- Runtime-Issued Access: Runtime-issued access is a pattern where credentials are created only when a verified workload requests them and expire when the task ends. The identity proves itself at request time, and the system issues a short-lived credential instead of storing a reusable secret. This reduces persistence and narrows abuse opportunities.
- Static Credential: A static credential is a long-lived secret such as an API key, password, token, or certificate that exists outside the moment of use. It creates persistent attack surface because it can be copied, stored, reused, and exposed across code, pipelines, configuration files, and third-party environments.
What's in the full article
Akeyless's full article covers the operational detail this post intentionally leaves for the source:
- How Gartner’s two April notes map to specific workload access and multi-vault governance categories
- The distinction between workload identity management, workload access management, and authorization management
- Why the CeDeSec pattern matters for centralized governance and decentralized enforcement
- How Akeyless describes its runtime access and governance model in relation to the Gartner reference architecture
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 June 23, 2026.
Updated on October 7, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org