Module 1

Introduction, overview, and threat modeling

The foundation for the track: Redwood Research's "AI Control: Improving Safety Despite Intentional Subversion" paper rendered in full with guided exercises woven in, then rebuilt as an interactive, calibrated demo you can play with.

1. The case for ensuring that powerful AIs are controlled Paper 105m
2. AI Control: Improving Safety Despite Intentional Subversion Paper 94m
3. Catching AIs red-handed Paper 30m
4. Prioritizing threats for AI control Paper 15m
5. How can we solve diffuse threats like research sabotage with AI control? Paper 12m

Introduction, overview, and threat modeling

Contents

Introduction, overview, and threat modeling

Contents