How is AI suddenly finding zero days that survived decades of human fuzzing? In April 2026, Anthropic's Claude Mythos and OpenAI's ChatGPT 5.5 rewrote what's possible with a keyboard. Host Tova Dvorin and offensive security expert Adrian Culley unpack the UK AISI's verdict — genuine step changes — through decades-old bugs in OpenBSD and FFmpeg, the 32-step autonomous attack chain, the Cyberstrike AI campaign across 648 firewalls, and why "capability doubling every 4.7 months" makes continuous validation non-negotiable for every CISO.

Rss Apple Podcaster
Podden och tillhörande omslagsbild på den här sidan tillhör SafeBreach. Innehållet i podden är skapat av SafeBreach och inte av, eller tillsammans med, Poddtoppen.