The role of emerging tech and AI in cyber security and defense tech
70 views
Jun 19, 2025
A panel of C4 and cyber tech experts talks about how emerging AI capability is broadening frontiers in defense… and what comes next.
View Video Transcript
0:00
Welcome back one and all. Next up, I would like to introduce our panel of experts this time now for a discussion on emerging technology as well as the role of artificial intelligence in both cybersecurity as well as defense AI. What a crucial topic. My guests for this session are Jonathan Elliott. He is the acting director of the AI rapid capabilities cell at CDAO. Matt Turek is here. He's deputy director of the information innovation office at DARPA and also Rachel Bondi
0:27
Ms. Bondi is the technical director of the Cybersecurity Program Office for PEOC4I
0:33
Mr. Elliott, Mr. Turek, Ms. Bondi, it is great to have each and every one of you to join us here for our second event today
0:39
Thank you all so much. Happy to be here. All right, awesome
0:44
Ms. Bondi, let me kick things off with you. Talk to me about one place where you think AI is already making an active difference today and one frontier maybe you would like to see it pushed a little further into the future
0:55
Yes, for us, I came in from industry, so I don't have a federal or military background
1:03
I was working with Eric Schmidt on Geosphere and Vint Cerf on DeepMind over at Google and then at IBM with Grady Broach and John Kelly on Watson
1:14
So when I arrived here, I was taking a look at how military systems were not prepared for being data centric
1:24
and so much of the AI solution set is data centric. We have been able to actually deploy AI cybersecurity
1:36
within our new systems for Zero Trust. And it working today At first we started on shore then we worked it into ships then we worked it into unmanned systems And it allows us to take a look at lane attacks and transversing
1:53
network attacks in a way that we weren't able to with traditional security. What I would like to
2:01
see more of is for us to continue to shift from network centricity to data centricity
2:08
Okay, I appreciate that. Mr. Elliott, from where you sit at CDAO's AI Rapid Capabilities Cell, I wonder what makes a capability rapid enough to actually earn the title of rapid? How does that sort of pacing look like from where you're positioned
2:24
Yeah. So I think the arc was created this past December and January to focus on Frontier AI and pushing forward the adoption of Frontier AI and generative AI technologies into the DoD
2:40
So the rapid is more a hallmark to our mission to like make sure that the DoD is does not lag behind where we've seen industry adopt and use successfully a lot of these generative technologies
2:54
And I think there are many examples of of use and benefit to the DoD
2:59
Just one is, as we've seen with industry, general use of tools like generative AI chatbots like chat GPT, Claude is examples
3:10
where you can now ask anybody at their desk can now have this tool there to help them research
3:15
summarize information, write for them. You see massive productivity gains from that
3:20
as well as also for scenarios like helping with cybersecurity or writing code and doing code debugging or processing the large amounts of information that the DOD receives that we can process now because we manpower limited and help making our warfighters more effective by taking that deluge
3:41
and summarizing it down into actionable movements forward. Yeah, I appreciate that
3:47
Mr. Turek, I want to bring you into this conversation. Grateful for you as well and, of course, your work at DARPA
3:51
how are you actively balancing high-risk experimentation with real-world application at DARPA? What does that kind of careful balance look like for you
4:01
Yeah, great question, and thanks for the opportunity to be here. Just to level set a
4:07
little bit, DARPA, Defense Advanced Research Projects Agency, we have the singular mission
4:12
of preventing or creating strategic surprise. That's the lens through which we look at any
4:17
potential investment that we make in things like AI and cyber. One of the things that we're doing to
4:23
rapidly move from transformative research capabilities all the way through into operational
4:30
use is an effort that we call Constellation, which is actually a partnership with Cybercom
4:34
that allows us to identify early stage transformational research capabilities, package them up in a way that might be operationally useful for cybercom, agree upon evaluation metrics
4:49
run those experiments to either support or refute the performance capabilities, and use that as a mechanism to span the valley of death and then hand them off to cybercom
5:00
Those agreed upon metrics are really the trigger for cybercom continuing to invest in that technology
5:06
to take it the final mile to integrate it into operational systems
5:09
and then to use it operationally. So sometimes DARPA innovations aren just the technology they the new processes and Constellation is a good example of that Ms Bondi I want to kick the conversation back to you and I really focused on what
5:23
it looks like to collaborate across agencies. So I'll phrase a version of this question, I believe, to all of you
5:28
But Ms. Bondi, how are you collaborating across different agencies, maybe specifically with
5:33
some commercial players, knowing that you need to accelerate smart, secure artificial
5:37
intelligence, but you need to do it in an effective way. And, of course, you need to do it in a way that doesn't really completely reinvent the wheel
5:45
Yes. For us, we're standing up a rapid capability office. And we're working with outside partners, DIU, AFWERX, and others
5:54
I recently was in South by Southwest talks in Austin about that exact thing, about how we could kind of crowdsource our efforts
6:04
And I think for me, it's dividing the mission set. So we have securing the enterprise, surviving the war, and then what I call stability or GRC, compliance use cases for cybersecurity
6:18
And understanding which is the mission focus for us is how we're approaching the human in the loop piece of working with AI to accelerate each of those mission sets with those partners
6:33
Commercial technologies is our aim. So I'm working with different providers on not only large language models, but small language models for specific use cases
6:46
And as we mentioned, data is just so there are so many pieces of data for cybersecurity and cybersecurity alerts that we really do need to use this in a human centric point of view to help our operators, even if they're operating unmanned craft
#news