I'll be at COLM 2026 to present self-incrimination training. After over ~7 months, I still think that self-reporting is a promising direction for monitorability in many ways.
Lmk if chatting with me will be helpful for your research!
I work on monitorability and AI Control.
Can we catch misaligned agents by training a reflex that fires when they misbehave? A simple impulse can be easier to instill than alignment and more reliable than blackbox monitoring.
We introduce Self-Incrimination, a new AI Control approach that outperforms blackbox monitors
