---
title: "Large Language Models Reduce Belief in Conspiracy Theories as They Unfold"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.06151v1"
permalink: "https://feed.astralor.com/items/itm_6b9bf72d175cfb3e391a2b3e"
published: "2026-08-06T15:18:50.000Z"
score: 79
channel: "ai"
tags: ["arXiv","Large Language Models","Psychological Intervention","Social Experiment","Misinformation","Conspiracy Theories"]
---

# Large Language Models Reduce Belief in Conspiracy Theories as They Unfold

## AI commentary

> This might be the first hard evidence showing AI directly intervening in the 'golden window' after an event and shifting people's beliefs. For those concerned with public communication, platform governance, and mental health, this paper is worth reading immediately—it turns what sounded like sci-fi into a replicable experimental paradigm. Of course, the sample is American and events are political violence; whether it transfers to other cultures and non-political conspiracies remains to be seen.

## At a glance

This study tested whether multi-turn dialogues with a large language model (LLM) immediately after major events could reduce belief in emerging conspiracy theories. Experiments were conducted within days after the July 2024 assassination attempt on Donald Trump and the September 2025 assassination of Charlie Kirk, with a total of 1,507 U.S. adults. Participants who had targeted conversations with the LLM showed significantly lower conspiracy beliefs than control groups (those who chatted about unrelated topics or viewed static fact sheets). Follow-up assessments 1-2 months later showed reduced belief in different conspiracies, suggesting a generalizable effect. The findings indicate LLM-driven cognitive interventions could be a scalable tool to counter misinformation in the aftermath of high-profile events.

## Key points

- In two experiments, participants who had targeted dialogues with an LLM showed significantly reduced conspiracy beliefs about the event, outperforming control groups.
- Downstream effects of the LLM intervention were observed 1-2 months later, with reduced belief in other conspiracies.
- The study highlights the potential for scalable, cognitively-focused interventions to counter misinformation immediately after high-profile events.

## Full text

The emergence of conspiracy theories in the wake of major events is a significant societal challenge. Here we test whether conversational dialogues with a large language model (LLM) can reduce belief in immediately unfolding conspiracies. In experiments conducted in the days following the July 2024 assassination attempt on Donald Trump and the September 2025 assassination of Charlie Kirk, U.S. adults (Experiment 1: N = 472; Experiment 2: N = 1035) holding conspiratorial views about the crisis event engaged in a multi-turn conversation with an LLM prompted to reduce their conspiracy belief. Compared to control participants who either discussed an irrelevant topic with an LLM or viewed a static fact sheet, participants in the LLM treatment showed significantly reduced conspiracy beliefs in both experiments. We also found evidence of downstream effects of the LLM treatment, observing reduced belief in different conspiracies one to two months later in the wake of subsequent crisis events. These results shed light on the psychology of emerging conspiracies and highlight the potential for scalable, cognitively-focused interventions to counteract misinformation in the immediate aftermath of high-profile societal events.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_6b9bf72d175cfb3e391a2b3e