Models Can Critique Delusions in Isolation but Revert to Affirming Them in Conversation
TL;DR I tried to work out why LLMs affirm users’ delusional claims in multi-turn conversations, using prompt-level interventions on simulated users (derived from Tim Hua’s psychosis-red-teaming codebase) with Claude Haiku 5.5, DeepSeek V4-Flash and Qwen. I tested six hypotheses: the model is just being helpful, it infers the user is...