Chris_Leong comments on Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training

[Chris_Leong]

Do you have any thoughts on what kind of experiments you’d like to see people run that would be more directly analogous?