The National Science Foundation (NSF) is using zero trust principles to prepare its data for artificial intelligence (AI), Michael Hauck, the agency’s acting chief data officer (CDO) and acting chief AI officer, said today.
Speaking at the ServiceNow 2026 Government Forum in National Harbor, Md., Hauck said that the agency is updating its architecture to support AI deployment across the organization.
“I have to be frank, at this point in time, most of our data is not really AI-ready,” Hauck said. “We’ve been a data-centric organization for … many decades … We’ve been in the business so long, our architecture is circa 19, so we need to upgrade that architecture in order to help us make our data AI-ready and also make our processes AI ready.”
As NSF modernizes its systems – connecting enterprise platforms with cloud collaboration and storage tools through application programming interfaces – the agency is running into new data security and governance challenges.
Hauck explained that integrations allow NSF’s enterprise platform, ServiceNow, to handle user interactions while data resides in Microsoft platforms. However, the agency still lacks full visibility into how those systems work under the hood and how to control data flows between them.
“How do we ensure that this application gets information from this one when it needs it, but doesn’t get information it’s not supposed to have?” Hauck said.
Zero trust – specifically its data pillar – has been key to addressing those challenges, Hauck said. He explained that using comprehensive data catalogs and standardized governance tags can help control how information flows across systems.
Notably, those data tags must align with identity and access management permissions so that each protected dataset has a corresponding entitlement or “key,” Hauck added.
The importance of data tagging has also recently drawn attention from the Federal CDO Council and the Chief Information Officer Council, he explained.
“There’s all this realization that data tags are important, yet we lack a standard that everybody can employ for those data tags. So that’s the thing that we need to work on,” Hauck said.
Patrick McGarry, federal chief data officer at ServiceNow, pointed to the importance of culture in upholding those principles.
“You can have all the best technology in the world, but if you don’t have the culture to go with it, it won’t matter,” McGarry said.
Hauck agreed and said that building data awareness beyond top technical leadership is key to preparing the federal workforce for AI.
“There’s no ‘I’ in data, but there is in ‘AI,’ so we need to focus on the intelligence part,” Hauck said. “And which intelligence part in AI we need to focus on, I think, is the human intelligence part in understanding the relationship between data and AI.”
“Development of the workforce is critical. Training resources are critical, and also how we communicate this out among ourselves, within the councils, but then also within our staffs and then within our organizations,” Hauck said.