Click here to watch the full discussion.
This June will be seven years since the Office of Management and Budget issued the first federal data strategy.
Just last March, it was the two-year anniversary of the first governmentwide policy around artificial intelligence.
Both of these firsts marked a shift in how agencies were to use data to improve mission outcomes.
By having a modern data strategy, agencies were taking a key first step to prepare for applying AI tools and capabilities. Agencies had to get their arms around things like data governance and data management before even testing the AI waters.
While managing data is always a challenge given the volume and variety that organizations face, many agencies over the last seven years have made enough progress to feel comfortable with starting to pilot AI tools for everything from automating mundane tasks to analyzing mission-critical information to interacting directly with citizens.
As the use of AI capabilities increases, agencies now have to be confident in the security and trustworthiness of these tools.
Not surprising, experts say having quality and authoritative data is the main driver of an organization’s success using AI.
“With the advancements in AI, we have recognized that there needs to be a data governance specifically for the National Nuclear Security Administration that talks about the different types of data at the classifications that we have,” Karen Sutton, the chief technology officer at the NNSA in the Energy Department, said on the discussion Mission-Ready AI: Where Data Strategy Meets Real-World Impact. “What we’ve done is for several years now at DOE, led by one of the NNSA organizations, is to hold DOE data days. We’ve seen over the last few years how that event has gone from sharing like 4 to 10 different groups to this year, they had to turn people away. We’re working on data problems and challenges, which is fantastic, and we’re trying to help us evolve the strategy that we use.”
The DOE data days also bring forward the recognition that as the use of AI increases so does the volume of data across the department. Sutton said that realization is forcing the agency to focus on data quality, having curated data and the ability to share data across the different program offices, different labs, plants and sites within NNSA.
“The intent of it is to really talk about data challenges, and to talk about processes and projects that are happening across the department so that groups can leverage what each other are doing to ensure that from an enterprise perspective, we have an approach that will allow us to share data across, which is different than has actually happened in the past,” she said. “We also have a significant amount of data that goes across both classified and unclassified environments. We have to have a data strategy that would work in multiple locations for multiple networks. Our focus has to be more on making sure that we’re applying the best practices to ensure we have rigorous data governance, effective management of our systems and our plans, while also making sure that we’re controlling the quality of the data because we need to ensure that if decisions are being made on that data, that the decisions are being made on the best and most effective data that we have.”
Moving to be strategic partner
Managing data in huge volumes is major challenge for the Postal Service’s Inspector General Office.
Berivan Demir Neubert, the director of analytics operations and governance for the USPS OIG, said the office brings in 110 petabytes of data every year, and they have to be selective about the data that they bring in house.
“We have to make sure that it’s supporting our mission to promote the integrity, accountability and efficiency of the U.S. Postal Service and its regulator,” Neubert said. “Once we moved into a more modern data system, we then migrated it into a better governed system as well. We are continuously improving and trying to figure out what’s next. Our goal in the next year is really maturing from being a data provider for our agency, where we’re just bringing in data to meet the analytic needs or the ad hoc requests that we’re getting from our investigators and our auditors, to being more of a strategic partner.”
Part of being a strategic partner is protecting the data with role-based and attribute-based access controls. Neubert said this becomes especially important as generative AI tools become more common and easy to use.
“We have a lot of application programming interface (API) endpoints connecting to private company data to really enhance what we’re working with to bring value to our agents and our auditors,” she said. “We have taken in the past few years the time to really tag documents and get our metadata populated in a way that is robust and is taking into account all of the data that we are housing within the OIG.”
By wrapping data in security, governance and controls, agencies are creating a data culture for everyone from data owners to AI developers to data scientists.
AI’s delicate balance
Brent Hansen, the chief technology officer at Optiv + Clearshark, said putting the data oversight and transparency in place is how agencies can speed up the adoption of AI and drive innovation into the mission areas.
“What that does is put a lot of emphasis on going back to the basics, which is the data is the most invaluable thing that we have. It’s making us go back and do the curation to understand the pipelines and to really marry the data owners with data scientists and those that are doing DevSecOps and creating these pipelines,” Hansen said. “When you look at having that level of maturity around governance, you can really move a lot faster. When you can understand who can see the data, what systems can see the data and what data can mix, there’s a lot of goodness there.”
One big lesson all three experts offered was the idea of slowing down to speed up.
Hansen said sometimes doing the hard work around data governance, classification and tagging will let agencies bring in AI tools and expand their use much more quickly than otherwise.
“The entire AI effort is incredibly delicate. You have to secure the data, move with caution, move with guardrails and move in harmony with your chief information officer, your security department, your data owners by ensuring everyone is on board and really being good stewards of not only the data, but of cybersecurity itself,” he said. “When you’ve got harmony and when you have frameworks, that is the discipline and that is the AI and data bible that everyone needs to ensure that you’re doing things from a governance side and also doing it from a cybersecurity perspective. Those are important because we do have to move forward responsibly, but make no doubt we can move faster than ever imagined today, than even just a couple of years ago.”
Copyright
© 2026 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.
