How Can You Tell if You've Instilled a False Belief in Your LLM?
Shows that common LLM belief metrics can be satisfied by simply prompting a model to pretend a false fact is true, and proposes harmfulness-based tests to separate real belief from role-play in safety experiments.