Threat modeling life: Prepping for the rest of us with Michal Zalewski (lcamtuf)

Hidden bunkers, stacks of canned food and piles of artillery. Disaster preparedness has become an Internet meme and these are some of the “prepper” community’s showcase images. But most of us who have lived through the recent pandemic, the Capital insurrection on January 6th and more no longer take the threat of a major disaster lightly. For those of us not willing or able to dig out a backyard bunker, is there a rational middleground where we can feel well-prepared for whatever comes next?

Software security legend Michal Zalewski (lcamtuf) answers this question and many others in his third book Practical Doomsday: A User's Guide to the End of the World. Using familiar threat modeling principles, Michal explores everything from evacuation gear and bulletproof vests to the genuine probabilities of civil war and a zombie apocalypse. In what can only be described as an unbelievable coincidence, Jack and Dave’s hour long interview with Michal was recorded the same day Silicon Valley Bank collapsed and was taken into government receivership.

In spite of the understandably dire subject matter, Michal’s equal sense of optimism and pragmatism steer us towards the middle path of rational risks and what a “normal” person should consider doing to be ready. It’s not nearly as hard as you might think and the peace of mind gained was well worth taking a hard look at the worst case scenario.

This interview is nearly cleanly separated into two parts as we focus on the opportunity and threat of artificial intelligence around the 32 minute mark, starting with Michal’s approach to writing. The real threat of generative AI to drive truly deceptive attacks takes center stage as we explore how the ability to easily generate compelling documents, images, video, etc. may make it nearly impossible to distinguish between reality and a scam.

No conversation on AI and threats seems to be able to avoid mention of the singularity threat, however, Michal keeps true to form and narrows in on the much more likely “paperclip problem” of mundane AI optimizing humans out of existence. This was one of our favorite episodes in ages, we hope you enjoy it and learn as much from it as we did. We also hope you got your money out of SVB, just like Dave did the week after this was recorded. Stay safe.