NIST publishes design tenets for AI agent identity and authorization

NIST published the second installment of its Cybersecurity Insights blog series on agentic identity, authentication and authorization on October 27, 2026, proposing an initial set of ten design tenets for agent identity management systems.

The tenets flow from public comments on an NCCoE concept paper, titled Accelerating the Adoption of Software and Artificial Intelligence Agent Identity and Authorization, published February 5, 2026, with comments due April 2, 2026, which drew feedback from more than 600 commenters across industry, government and academia.

What it is

NIST's National Cybersecurity Center of Excellence said it is interested in launching a project to demonstrate how identity standards and best practices can be applied to software agents, with a focus on agentic AI applications.

Following the comment period, NIST said its Secure Software Development (DevSecOps) project will provide the first implementation use case for this project to demonstrate how AI agents can be identified, authenticated, and authorized within the software development lifecycle.

This identity work sits alongside NIST's wider research agenda, where the agency says it conducts fundamental research into agent authentication and identity infrastructure to enable secure human-agent and multi-agent interactions.

What it requires

The design tenets call for agentic infrastructure to enable the unique, cryptographically verifiable identification of agents distinguishable from other identities such as human users and service accounts.

Access grants are expected to be Just-In-Time, time-bounded and audience-specific rather than relying on long-lived bearer tokens.

The tenets also require that the logic that determines what an agent attempts to do is architecturally separated from the systems that determine what it is allowed to do, because large language models accept data and instruction as a single stream of tokens through a single interface.

What security leaders should do

Security leaders should begin treating AI agents as distinct identities with their own credential lifecycles rather than as extensions of human accounts, since that is the direction NIST's design tenets point. Teams can use the ten tenets as a checklist when evaluating vendor claims about agent identity, authentication and authorization, particularly around token lifetime, scoping and human-in-the-loop escalation. Because the NCCoE project is still testing a DevSecOps implementation and collecting feedback on a rolling basis, organizations should track NIST's blog series and resource hub before locking in long-term agent architecture decisions. Security teams piloting agentic tools now should favor short-lived, scoped credentials over shared or static API keys wherever current tooling allows it, and document where that is not yet possible.