NIST targets AI agent identity as perimeter model fades
The NCCoE's six AI-cyber projects and its agent identity concept paper are building the governance framework for a perimeter-less world that most agencies still haven't acknowledged they're living in.
TL;DR
NIST's NCCoE is running six projects at the AI-cybersecurity intersection, and the agency's February concept paper on software and AI agent identity and authorization laid out the scaffolding for governing what AI agents can access and do. The argument that perimeter-based federal security is obsolete isn't new, but NIST's standards work confirms the government's own technical body is building the replacement. The gap isn't diagnosis; it's whether agency CISOs have moved their architectures to match.
The federal cybersecurity perimeter was always a convenient fiction. Useful for budgeting and architecture diagrams, less so for actual defense. But AI agents have made it untenable in a way that even the most committed perimeter-defenders can't ignore. When an AI assistant embedded in an already-approved productivity suite can summarize, analyze, and exfiltrate sensitive data without crossing a network boundary, the boundary stops being the right place to put your controls.
NIST saw this coming. The NCCoE released a concept paper in February on software and AI agent identity and authorization, asking how existing identity standards could apply to autonomous agents that act at machine speed without the governance historically applied to human users. By June, NCCoE director Cherilyn Pascoe confirmed that AI was "going to be part, if not a leading part, of every project going forward at the center," with six distinct AI-cyber projects running simultaneously, including a "Cyber AI Profile" and work on securing agentic AI. The concept paper asked a question that is now operational for every federal CISO: how do you apply identification and authorization controls to something that isn't a person?
The FNN commentary's recommendation that agencies govern identity, along with the data and compute resources those identities access, is the right prescription. But it's also the prescription NIST is already writing. The standards body has moved from diagnosis to draft guidance. Most agencies haven't. For the security engineer, that means identity governance applied to both human users and the AI agents they deploy. Continuous evaluation of access, driven by behavior and device posture, has to become the default whether or not the agency's policy framework has caught up.
Published ·Deep Fathom