In the third and final installation of our podcast miniseries on artificial intelligence, produced in association with Darktrace, we delve into the realm of AI fighting AI—or what researchers refer to as “adversarial AI.”
Adversarial artificial intelligence can take many forms—as a tool for hacking through AI-powered security of other systems, for example, or deceiving another algorithm with input that causes a specific, fake result. Ars editors Sean Gallagher and Lee Hutchinson spoke with the leader of the winning team from the 2016 Defense Advanced Research Projects Agency’s Cyber Grand Challenge, ForAllSecure CEO David Brumley, about advancements in AI-driven hacking. Lujo Bauer, a professor of electrical and computer engineering at the Institute for Software Research at Carnegie Mellon, joined Lee and Sean to talk about his research into ways to use AI to defeat technologies such as facial recognition. And Max Heinemeyer, director of threat hunting at Darktrace, discussed research already being done into how to stop AI-driven attacks on computer networks.
This special edition of the Ars Technicast podcast can be accessed in the following places:
https://itunes.apple.com/us/podcast/the-ars-technicast/id522504024?mt=2 (Might take several hours after publication to appear.)