A top official at the Department of Homeland Security (DHS) said today that he is worried about an uptick in the adversarial use of artificial intelligence technologies, which have the potential to cause harms to U.S. critical infrastructure.
At the Dell Technologies Forum in Washington, D.C., today, Dimitri Kusnezov, under secretary for the Science and Technology Directorate (S&T) at DHS, said that adversaries can leverage AI not only for cyberattacks but also to undermine democracy.
“I worry a lot about the adversarial side of AI,” Kusnezov said. “It is beyond a cyber problem as we worry about things like a transnational repression of undermining democratic institutions.”
Just last month, DHS released its annual Homeland Threat Assessment report for 2024, which said adversaries will leverage AI to create more believable misinformation campaigns and use the emerging technology to develop more evasive cyberattacks on U.S. critical infrastructure.
For example, Kusnezov said DHS is worried about adversarial use of deepfakes – which use a form of AI called deep learning to create images, videos, or audio recordings of fake events.
He explained that this could pose a problem for DHS’ Citizenship and Immigration Services component, which might not know if they are interviewing someone who is real or not.
Additionally, he warned that first responders and 911 call centers could get inundated with “fake SWAT team requests,” where an automated voice could falsely claim to be hiding in an elementary school from a school shooter and answer questions in real-time.
“They’re done with voice over the internet, so you can’t track these,” Kusnezov said. “So, there’s this side of things that is different than kind of the cyberattacks, but is real. And there’s also the enhancement of cyberattacks and the positive and negative consequences, especially to our critical infrastructure in the U.S.”
Kusnezov co-chairs the DHS Artificial Intelligence Task Force – which Secretary Alejandro Mayorkas established in April – alongside DHS Chief Information Officer (CIO) Eric Hysen.
As DHS is thinking about these very real threats, Kusnezov said that the task force also is looking to report on future AI opportunities. But first, he said his agency conducted an assessment phase to see where the agency was in its AI journey and gain a better understanding of responsible AI.
“We spent the first part of the first phase of this talking with all our components and getting from them a sense of where things are, and a sense of where they want to go,” he said. “And the next phase – which we’re entering in now – is surfacing the larger opportunities that we see are going to be important, and we’ll eventually report out early summer.”
“We’re excited to hear what comes out of it, and what we’ve gleaned from all this information,” concluded Surya Durvasula, vice president of Federal Civilian Sales at Dell Technologies.
