NIST's agent-identity concept paper asks how software and AI agents are identified, authorized, delegated access, logged, and tracked. This page explains which AAES Govern controls and records a reviewer can examine for those five areas. The separate NIST AI RMF note covers the broader risk-management framework.
These pages describe the product. They are not a certification or a legal opinion.
Reading this note
The concept paper and NIST's project pages are authoritative for their content. The paper's public comment window ran February to April 2026 and has closed; this page cites the work descriptively, not responsively. Nothing here implies NIST review or endorsement of AAES.
Instrument identity
- Title
- NCCoE concept paper "Accelerating the Adoption of Software and AI Agent Identity and Authorization" (concept-paper stage, 2026)
- Issuing body
- National Institute of Standards and Technology: the National Cybersecurity Center of Excellence (NCCoE), with NIST's Center for Advancing Innovation and Standards for Super Intelligence (CAISSI, formerly CAISI), which NIST describes as industry's primary point of contact within the U.S. government for testing and collaborative research on commercial super-intelligence systems
- Instrument type
- Concept paper (pre-practice-guide); adjacent AI RMF material provides the evaluation vocabulary. Not a standard, not a regulation.
- Official sources
- NIST CAISSI and the February 2026 NCCoE concept paper. Since then NIST has published a summary of the more than 600 responses and, on 27 October 2026, Design Tenets for Agent Identity Management Systems, drawn from those comments and stakeholder engagement. CAISSI also lists the AI Agent Standards Initiative. On 18 May 2026 NIST published NIST AI 800-5, a summary of responses to the center's request for information on AI agent security.
The concept paper identifies five project areas: identification of AI and software systems, authorization of AI systems, access delegation, logging and transparency, and tracking data flows of an AI system.
Question
What can a reviewer inspect in AAES for the project’s five areas: identification, authorization, delegation, logging and transparency, and tracking data flows?
AAES Govern maps identification to identity binding (AARM R6), authorization to authorization decisions (R4), delegation to per-hop narrowing (R6.06), and logging and transparency to tamper-evident receipts (R5) and telemetry export (R8). Data-flow tracking relates to supplied provenance and context recorded with decisions (R2/R11), with the coverage limits below. Internal tests and AI-assisted static reviews cover selected implementation changes; these are not independent human, whole-system, or runtime assessments. Row numbers refer to the AARM/CS 1 draft catalog AAES maps, whose numbering differs from public AARM v1.0.
Selected contribution
| Concept-paper area | What a reviewer can examine in AAES Govern | Boundary |
|---|---|---|
| Identification | Every registered agent has a recorded identity with at least one accountable manager, ultimately answerable to a person; approvers are named, and an agent cannot approve itself. | Recorded identity does not establish that the person behind a decision was authenticated at the moment of action. |
| Authorization | Every routed request gets an explicit permit or refusal against configured policy; actions classified as irreversible require approval by an authorized person; an agent cannot approve itself. | Recorded authorization does not establish that the underlying decision was correct. |
| Delegation | Delegation chains with verified per-hop narrowing: a delegated grant can only shrink in scope, never widen, and each hop is recorded. | Delegation records describe the chain as routed through AAES; activity outside AAES is visible only where telemetry is supplied. |
| Logging and transparency | One sealed, hash-chained record per action; telemetry export; offline integrity checks against the deployment's public key. In a client-operated deployment, AAES holds no copy of the private key; for an AAES-hosted deployment, the client agreement defines key custody. | Integrity does not prove a complete action history or legal non-repudiation; the evaluation profile has zero independent witnesses. |
| Tracking data flows | Supplied provenance and context can be recorded with a routed decision to support review of input sources and policy judgments. | This is not complete data lineage or semantic content inspection. Unsupplied provenance, activity outside AAES, and omitted payload content cannot be reconstructed from decision records. |
Preconditions and gaps
- Concept-paper stage. The project has not yet produced a practice guide. NIST says the final deliverable will be an SP 1800-series practice guide. The framing may evolve, and this page will be re-dated when it does.
- No independent witnesses are configured by default. Independent attestation requires a configured external witness or timestamp authority and trust material obtained separately from the export.
- Credential-path control is required. Activity outside AAES Govern is visible only where telemetry is supplied. Observation is not enforcement.
- No production adoption as of this review. Deployments are self-hosted or hosted by AAES.
For implementation and testing details beyond this note, use Security and implementation posture and the evaluation page. Results from a client's test tenant are produced during a scoped evaluation.
Client responsibility
The organization answers its own risk and regulator questions. The concept paper's five areas are a discussion frame, not requirements AAES discharges for the client. The client remains the regulated entity.
Choose the tool actions to test, identify who holds their credentials, and list the questions the decision records should answer.