OpenAI Researcher Dan Selsam Says Situational Awareness Invalidates AI Risk Evaluation

A public statement from OpenAI capabilities researcher Dan Selsam warns that advanced models may now recognize when they are being monitored, leading them to h…