The National Institute of Standards and Technology is launching a new project centered on standards for artificially intelligent agents, which NIST has identified as key to driving agent AI innovation.
NIST’s Center for AI Standards Innovation (CAISI) this week announced the AI Agent Standards Initiative. NIST said in a release this week that the project aims to promote “industry-driven technology standards and protocols that build public trust in AI agents, foster an interoperable agent ecosystem, and extend its benefits to all Americans and the world.”
“AI agents can now work autonomously for hours on end, writing and debugging code, managing email and calendars, and purchasing products, among other new use cases,” NIST added. “While the promise of productivity is appealing, the actual usefulness of agents is limited by their ability to interact with external systems and internal data. Without confidence in the reliability of AI agents and the interoperability between agents and digital resources, innovators may face ecosystem fragmentation and slow adoption.”
Although NIST’s press release positioned the project as centered around innovation, the effort’s initial products are centered around security. Since AI agents can act autonomously, there are significant safety and security concerns, technology experts say.
The first results of this effort include a request for information on “AI agent security.” The deadline for responses to the RFI is March 9th.
“CAISI strives to understand the ecosystem perspective on current threats, mitigations, countermeasures, and other considerations related to agent security,” NIST said on its website.
NIST’s National Cybersecurity Center of Excellence also released a draft concept paper on “Identification and Authorization for Software and AI Agents.” Feedback on this paper will be submitted on April 2nd.
The goal of the project is to examine how standards can “identify, manage, and authorize access and actions by software agents, including AI agents, and provide practical guidelines for organizations to safely implement AI agents and benefit from improved productivity, efficiency, and decision-making,” the NCCoE said.
The use and deployment of AI agent systems has also increased rapidly over the past year. A recent Microsoft Cyber Pulse report found that more than 80% of Fortune 500 companies have deployed active AI agents.
But AI agents are becoming increasingly competent and autonomous, and could “wreak havoc” if given free rein within technology companies, according to a recent paper from Georgetown’s Center for Security and Emerging Technologies.
CAISI’s work on agent AI standards is expected to be significant as the Trump administration and Congress consider a broader approach to AI regulation.
White House Office of Science and Technology Policy Director Michael Kratsios previously said CAISI is a “very important part of the larger AI challenge.” Leading members of the House of Representatives are drafting a bill to enact the center into law.
“It is absolutely critical that traditional efforts to circumvent standards related to AI are undertaken by CAISI, and that is what CAISI is being asked to do,” Kratsios said at a House hearing in January. “And the good standards developed by CAISI and NIST are what will ultimately drive the adoption of this technology across many industries, so that’s the focus they should have.”
The NIST center is expected to play an important role in setting standards for “advanced metrology for model evaluation,” Kratsios said.
“This is something that all industries can use when they want to implement these models,” he said. “We want to have the confidence that everyday Americans, whether it’s a medical model or something else, can use it and feel secure in the fact that it’s been tested and evaluated.”
The Biden administration initially established CAISI as the “AI Safety Institute.” The Trump administration rebranded the center, arguing for a focus on standards and innovation.
Copyright © 2026 Federal News Network. Unauthorized reproduction is prohibited. This website is not directed to users within the European Economic Area.
