The U.S. National Institute of Standards and Technology has launched an initiative focused on making AI agents more interoperable, secure and trustworthy.
AI agents can plan tasks, call software tools and exchange information with other systems. That flexibility makes them useful, but it also creates new risks: an agent may receive untrusted instructions, overreach its permissions or pass sensitive data across organizational boundaries.
NIST’s standards work matters because companies are building agent platforms with different protocols and security assumptions. Common technical foundations could make it easier to connect agents without forcing every organization to design its own identity, authorization and audit mechanisms from scratch.
The initiative also signals that agent security is moving from an experimental concern to an infrastructure requirement. Enterprises considering autonomous workflows will need clear controls over what an agent can access, which actions require human approval and how activity is recorded for review.
Source: NIST
