Approved routing is the practice of sending AI requests through sanctioned models, environments, or control layers based on policy. It helps keep sensitive prompts out of unmanaged systems while preserving usable workflows for employees and developers.
Expanded Definition
Approved routing is an AI governance control that determines which sanctioned model, workspace, gateway, or orchestration layer can receive a request before any prompt, file, or tool action is executed. It is used to keep enterprise data within designated boundaries, reduce exposure to unsanctioned AI services, and apply policy-based checks for sensitivity, user role, geography, or task type. In practice, approved routing is narrower than general access control because it focuses on NIST Cybersecurity Framework 2.0 style governance for AI usage paths, not just login events or network segments.
Definitions vary across vendors on whether approved routing is implemented as a proxy, an AI gateway policy, a browser mediation layer, or a workflow rule inside a platform. NHI Management Group treats the term as a control pattern rather than a single product feature. Its purpose is to constrain where AI interactions can occur, especially when employees use large language models, retrieval systems, or agentic tools that can invoke external services. The most common misapplication is treating approved routing as a simple URL blocklist, which occurs when organisations filter destinations but fail to govern sanctioned paths, context transfer, and tool execution rights.
Examples and Use Cases
Implementing approved routing rigorously often introduces latency and policy-maintenance overhead, requiring organisations to weigh tighter data control against faster user access and simpler developer workflows.
- An employee drafts a customer response in a company-approved AI assistant that routes prompts through an internal model endpoint rather than a public consumer chat service.
- A developer sends code review requests through a sanctioned orchestration layer that redacts secrets before forwarding context to a model provider.
- A regulated business routes prompts containing personal data only to an approved environment with logging, retention limits, and regional processing constraints.
- An agentic workflow uses an approved router to decide whether a task may reach external tools, or must remain inside an internal execution boundary.
- A security team pairs approved routing with policy enforcement to prevent sensitive prompts from being pasted into unmanaged browser extensions or shadow AI tools.
For identity-adjacent workflows, approved routing is strongest when it is tied to user assurance, device trust, and data classification rather than a blanket permission model. That is why many organisations map it alongside governance guidance from NIST AI Risk Management Framework and operational patterns described in OWASP guidance for LLM applications. The control is not about banning AI use; it is about ensuring the request reaches the right governed path for the right purpose.
Why It Matters for Security Teams
Approved routing matters because AI risk often begins before a model generates anything. If requests can reach unmanaged systems, security teams lose visibility over what data was exposed, where it was processed, and which downstream tools were activated. That creates governance gaps across privacy, secrets handling, retention, and incident response. For identity security teams, the term also connects to Non-Human Identity governance when agentic systems or service accounts are allowed to route requests on behalf of users, because the routing decision becomes part of the trust boundary.
Security leaders should treat approved routing as a policy enforcement point for AI usage, not a convenience layer. It supports segmentation between sanctioned and unsanctioned AI services, and it helps preserve auditability when employees use AI to handle sensitive content. The concept aligns with enterprise risk management in NIST Cybersecurity Framework 2.0 and with zero trust thinking when routing decisions depend on context rather than network location alone. Organisations typically encounter the need for approved routing only after a sensitive prompt has already been sent to an unmanaged model, at which point the control becomes operationally unavoidable to contain exposure.
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 and OWASP Non-Human Identity Top 10 address the attack and risk surface, while NIST CSF 2.0, NIST AI RMF and NIST SP 800-63 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| NIST CSF 2.0 | GV.OC-01 | Defines governance outcomes that fit policy-based AI request routing. |
| NIST AI RMF | AI RMF governs trustworthy AI use, including policy and oversight of routed requests. | |
| OWASP Agentic AI Top 10 | Agentic AI guidance addresses tool use and execution boundaries relevant to routing. | |
| OWASP Non-Human Identity Top 10 | NHI governance applies when service identities or agents trigger approved routing. | |
| NIST SP 800-63 | AAL2 | Identity assurance helps ensure the right user context is routed to AI services. |
Bind routing permissions to non-human identities and review their permitted AI destinations.
Related resources from NHI Mgmt Group
Deepen Your Knowledge
Reviewed and updated by the NHIMG editorial team on August 15, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org
