Guidance

NIST Agent Identity (draft, Feb 2026)

NIST NCCoE: Software and AI Agent Identity and Authorization

7 guidance propositions · 12 connected unified controls

Open NIST Agent Identity (draft, Feb 2026) in the map

Version: February 2026 draft concept paper
Published: 2026-02-05
Reviewed: 2026-09-10
Amendments: Initial concept paper; comments closed 2026-04-02; project reviewing comments as of 2026-09-10

Draft concept paper seeking input on a proposed NCCoE project. The seven selected topics below summarize questions and areas of exploration, not final requirements or a completed implementation guide. NIST-AGI identifiers are local catalog references, not NIST control numbers. The project focuses on enterprise agents; external agents from untrusted sources are outside its initial scope. The NIST AI Agent Standards Initiative, launched 2026-02-17, is a standards-development program rather than a certifiable standard: https://www.nist.gov/artificial-intelligence/ai-agent-standards-initiative

Guidance propositions

“Informs” records a guidance relationship. It does not claim that a unified control satisfies a mandatory requirement.

Agent Identity and Authorization Topics

Proposition	Unified control and mapping
NIST-AGI-01Distinct agent identities and identity boundaries Source Source pages: Concept paper pp. 4, 6: Identification; Areas of Interest	UC-ACCESS-06 — Manage unique identities and identifiers end to endInforms
NIST-AGI-02Agent authentication and credential lifecycle Source Source pages: Concept paper pp. 4, 7: Authentication; Relevant Standards and Guidelines	UC-ACCESS-08 — Manage and protect authenticators across their lifecycleInforms UC-ACCESS-10 — Authenticate devices and services before granting connectionsInforms
NIST-AGI-03Context-sensitive authorization and least privilege Source Source pages: Concept paper pp. 4, 6: Authorization; Areas of Interest	UC-ACCESS-03 — Enforce least privilege, need-to-know, and segregation of dutiesInforms UC-ACCESS-05 — Enforce approved authorizations for information and functionsInforms UC-AI-19 — Constrain agent actions and tool use to authorized scopeInforms
NIST-AGI-04Delegated authority and human accountability Source Source pages: Concept paper pp. 4, 6: Authorization; Access Delegation	UC-ACCESS-05 — Enforce approved authorizations for information and functionsInforms UC-AI-19 — Constrain agent actions and tool use to authorized scopeInforms
NIST-AGI-05Verifiable agent action logs and authorization traceability Source Source pages: Concept paper pp. 4, 6: Auditing and non-repudiation; Logging and Transparency	UC-LOG-01 — Log security-relevant events across all systemsInforms UC-LOG-03 — Protect audit logs and retain them for required periodsInforms UC-AI-08 — Log and monitor AI system behavior in operationInforms
NIST-AGI-06Prompt-injection prevention and limits on resulting harm Source Source pages: Concept paper p. 4: Prompt Injection prevention and mitigation	UC-AI-18 — Defend AI interfaces against adversarial input, injection, and endpoint abuseInforms UC-AI-19 — Constrain agent actions and tool use to authorized scopeInforms
NIST-AGI-07Prompt provenance and data-flow tracking Source Source pages: Concept paper p. 6: Tracking Data Flows of an AI System	UC-DATA-11 — Control data flows, leakage, and cross-border transfersInforms UC-AI-09 — Govern AI data quality, provenance, and preparationInforms