Personalized dialogues with AI reduce conspiracy beliefs mainly by supplying relevant, well-reasoned counterevidence, according to a preprint by Gordon Pennycook and two co-authors submitted on Nov. 3, 2026.[1] In two experiments with 2,193 U.S. adults recruited online, the authors removed possible causes of the effect one at a time.[1] The effect held when the AI disclosed its persuasive goal, halved its replies or sent a single tailored message; only telling it to persuade without factual evidence and arguments substantially weakened it, they wrote.[1]