Feds Weigh Generative AI Use in Cybersecurity, Data Analysis

Federal officials see generative AI helping better identify cybersecurity threats, sort through massive amounts of data at speed and equip the workforce with information for driving better decisions.

The National Institutes of Health (NIH) and its parent agency, the Department of Health and Human Services (HHS), are exploring the key benefits of using generative AI for analyzing vast data sets. During ATARC’s Optimizing Operations: Gen AI in Federal Agencies webinar, Nathan Hotaling, senior data scientist with the National Center for Advancing Translational Sciences at NIH, said generative AI has been critical when looking for electronic health record data.

“It makes that unstructured data actually searchable, understandable and instantly findable. Functionally, it means that you are going to be able to harness all of that unstructured information that sits inside of PDFs and paragraphs of text in places that before you needed a human to read through,” said Hotaling. “Generative AI helps us move to that next level and my group is working on analyzing physician notes and we also use it to help our scientists find the relevant publications to the questions that they’re asking.”

Conrad Bovell, branch chief of cybersecurity advisory and strategy at HHS, said generative AI is impacting clinical data and is playing a critical role in treatment for cancer patients.

“In a proof of concept study, researchers at NIH have developed an artificial intelligence tool that uses routine clinical data, such as from a simple blood test, to predict whether someone’s cancer will respond to immune checkpoint inhibitors, which is a type of immunotherapy drug that helps immune cells kill cancer cells. So, the machine learning model may help doctors determine if immunotherapy drugs are effective for treating a patient’s cancer,” said Bovell.

Cybersecurity and AI

Cybersecurity is also a critical use case for generative AI. Bovell said it can be a powerful tool for cyber criminals and nation-state threat actors. It is also a powerful tool for the cybersecurity teams responsible for mitigating threats in security operations centers (SOC) and security information management (SIM) systems.

“In SOCs, [generative AI] can identify patterns indicative of cyber threats such as malware, ransomware or unusual network traffic that might also include data from other existing traditional detection systems,” said Bovell. “Generative AI contributes to more sophisticated data analysis and anomaly detection in SIM systems, as well. By learning from historical security data, AI models can establish a baseline of normal network behavior and then flag deviations that may signify security incidents. It can successfully take fingerprints of normal operational activity occurring in your environment.”

AI Can Help Human Decision-Making

During the webinar, Chakib Chraibi, chief data scientist at the Commerce Department’s National Technical Information Service, said another benefit of generative AI is that it helps to decrease human bias.

“Generative AI can help us in our decision-making process and can be leveraged to reduce human bias,” said Chraibi. “The Department of Justice can use generative AI to assist in analyzing case law demographic data and historical sentencing information to provide us with objective insights for sentencing recommendation and potentially for a fair outcome. So, fairness is an important aspect of what we can get out of properly leveraging AI.”

Chraibi noted the potential impact of generative AI on government operations and service delivery.

“[Agencies] can leverage it to enhance efficiency and drive innovation,” said Chraibi. “We also have to be strategic in implementing generative AI. We must make sure that we ensure compliance, data protection and the responsible use in enhancing the trust in AI while also enhancing the productivity and security within federal agencies.”