Hey, it’s Marek! And Ivano!
Two firsts these past few days: a co-authored newsletter and a guest on the podcast. Both with Ivano Bongiovanni, General Manager of AusCERT.
Ivano framed the conversation with a question I’ve been chewing on since: when AI agents become the “social” component of socio-technical systems, what happens to the frameworks we built assuming the social part was human?
The written post (published Friday) lays out the argument: two “trinities”: one describing a new kind of attacker, the other describing why our systems can’t defend against it. In the conversation, we go further. Ivano connects agentic AI ecosystems to the safety science of industrial disasters. We debate whether a kill switch is a meaningful concept in the AI era. And he’s refreshingly honest: “If I had the solution, I probably wouldn’t be here talking in this podcast.”
Give it a listen/watch. It’s raw, unscripted, and it’s the kind of conversation I want to have more of. If you enjoyed this format experiment (a conversation among two people), let me know.
Stay curious! Ciao!
