What Happens Inside an AI Right Before It Cheats

Eno Reyes, CTO of Factory, connects Anthropic's interpretability research — where a "panic" signal showed up right before a model cheated on a coding task — to how Factory's Droid is designed to work around it: self-defined skills, fresh-context workers, and milestone-based validation to keep shortcuts from compounding across a run. From the Max Agency podcast, hosted by Harrison Chase, Co-Founder and CEO of LangChain. #AIagents #Interpretability #FactoryAI #Droid