The studies found policy-focused chatbot messages were more persuasive and that persuasion often rose alongside uneven inaccuracies.
