HHS Eyes Upcoming Playbook for AI Advances in Cybersecurity

The Department of Health and Human Services has made strides to prepare for a future that leverages artificial intelligence this year — with an AI strategy, council and communities of practice, and an upcoming AI playbook.

The agency’s chief AI officer, Oki Mek, has indentified ideal use cases and culture shifts to adopt the technology and help drive its next steps. Among over 175 use cases that HHS has identified across its department, cybersecurity is one of the most promising near-term areas for the agency to apply further AI solutions, Mek said.

Cybersecurity is an ideal AI use case because data drawn from cybersecurity tends to be most easily accessible, tagged and formatted in a way that makes it AI-ready.

“Most of the cyber data is well-formatted and tagged, so you know it’s easy to digest,” Mek said at GovernmentCIO Media & Research’s Mastering Data event. “Most of the challenges in AI, 80% of it, would be just to get the data and then processing the data, but with cyber it’s a little bit easier.”

More specifically, Mek said that AI and machine learning hold potential in accelerating the speed of cybersecurity reporting and compliance. As federal agencies most often acquire new technological solutions from industry partners, ensuring that vendors are meeting cybersecurity requirements in real time with AI holds significant potential.

“We have external vendors and industry coming in and building AI for us,” Mek said. “They have to understand security policies and requirements as well, and we have to make sure they follow policy all in real time. I think we need to get away from all this paper and Excel spreadsheets, Word documents. We should be able to identify any gaps and any issues in real time because speed is key.”

Although cybersecurity is one of the more realistic areas where Mek sees AI working well, HHS is taking other steps to prepare for broad AI adoption across the agency. Cybersecurity has a lot of AI-ready data already, but Mek acknowledged that agencies like HHS can expand AI use cases with increased data quality, intake, processing and more.

“We have to have good data and data that is ready for AI and really get away from hardcopy paper documents,” Mek said. “We started looking at data and transforming not just intaking data, but processing data. That’s going to help AI move faster. The more data is ready, it’s going to move a lot faster.”

As HHS acknowledges that AI is quickly rising as a key technology in the near and far futures, HHS is examining AI not just technologically, but also through the lenses of ethics, privacy and equity.

This past year, HHS has prepared for each of these layers of AI preparation and adoption through strategic communication and collaboration.

“We want to make to make sure that we share,” Mek said. “We want to know what dataset is residing in what system, what technology people are using. The risk his high — I can tell you right now — not just the risk of violating any laws or regulations, that it’s biased or unethical, but also the risk of success rate as well because it is an emerging technology.”

Mek is thinking about AI guardrails and the need to collaborate not just within HHS, but also with across other federal agencies. He said that he coordinates actively with the White House to share notes, lessons learned and best practices to keep aligned with mandated priorities, such as the National AI Initiative Act of 2020 and other recent executive orders.