Chilling Risks of AI Deepfakes and Cyberattacks, w/ Security Expert Christian Seifert
AI deepfakes, cyberattacks, and cybersecurity risks in the age of artificial intelligence.
Guest
Christian Seifert
Researcher in Residence, Forta
Christian Seifert is a cybersecurity researcher and was Researcher in Residence at the Forta network during the episode. He previously led cybersecurity work at Microsoft. His expertise includes online abuse, adversarial behavior, prompt-injection risks, and the ways AI can strengthen both attackers and defenders.
Watch on YouTube ▶About this conversation
You get a FaceTime from a loved one. They're in trouble. Big trouble. Their life is in danger. They need you to Venmo you $, they have no choice.
Except...it's not them you're FaceTiming with, it's an AI-enabled deepfake that can talk to you real-time.
Welcome to the fun world of AI phishing and cyberattacks!
Christian Seifert, a veteran cyber-security expert (he used to head up cyber-security at Microsoft, and now he's Researcher in Residence at the Forta network), joins the show to unpack these kinds of risks. How worried should you be? How grim will the future get?
We cover:
How AI makes it easier for personalized scams and fakes (2:30); why AI security needs to be baked into the tech and tools we use today (8:30); why “prompt injections” pose a risk and what we can do about it (18:00 ); how AI agents could cause mischief as a type of “super virus” (20:00); whether he prefers open-sourced or closed-sourced AI development for AI safety (25:00); why Christian is cautiously *optimistic* that the future will bring an advantage to AI safety defenders (26:30); and recommendations to the listener of how they can minimize AI risk (29:30); and a surprise twist that AI is making the jobs of cyber-security... paradoxically easier?
The Forta Network:
https://forta.org
Christian Seifert on Twitter/X:
https://twitter.com/cseifert