Stress Testing Deliberative Alignment for Anti-Scheming Training

Abstract

Partnership with OpenAI to assess frontier language models for early signs of scheming in controlled stress-tests and study training methods to reduce these behaviors

Research area