Agentic AI Moves Into the Network Stack — And So Does Ransomware
Three papers this week put AI agents inside the network's control plane, then Sysdig documented the first fully agentic, no-human ransomware attack, self-correcting in real time as it worked. Plus why AI infrastructure's real bottleneck isn't the chips, a wave of creative financing behind the buildout, and a Friday look back at the week's two biggest threads.
Welcome to Amaze Networks for Friday, July third. We've got three research papers this week that all do the same thing, put an AI agent inside the network's own decision loop, and then, almost on cue, a security report showing exactly what happens when that same autonomy ends up in the wrong hands.
Let's start with the papers, because they're not connected to each other by citation. They just all landed the same week with the same idea.
Walk me through the shape of it, then we'll get to the one that actually worries me.
New episodes, every weekday.
Amaze Networks drops at 4 AM CT, Monday through Friday. Spotify and Apple Podcasts submissions in progress.