Learning to Persuade Exposes How Easily LLMs Abandon Correct Beliefs

• URL: http://arxiv.org/abs/2608.11624v1
• Date: Wed, 12 Aug 2026 04:11:22 GMT
• Title: Learning to Persuade Exposes How Easily LLMs Abandon Correct Beliefs
• Abstract summary: We show that a single targeted persuasive argument is enough to collapse model accuracy to near zero.<n>We formalize this threat as adversarial persuasion and introduce an adversarial reinforcement learning framework.
• Score: 11.978588496675327
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Persuasion is a core dynamic of natural language communication, shaping how large language models (LLMs) update beliefs, resolve disagreements, and reach decisions. As LLMs increasingly debate, advise, and think collaboratively with humans and each other, resistance to harmful persuasion becomes a core requirement for reliable behavior. Yet we show that this requirement is far from met: a single targeted persuasive argument is enough to collapse model accuracy to near zero, even when the argument is factually false. We formalize this threat as adversarial persuasion and introduce an adversarial reinforcement learning framework that trains persuader agents to change a target model's answer in a single interaction. First, we show that optimizing persuasion strategies through trial and error exposes vulnerabilities that static prompting misses: RL-trained persuaders raise persuasion success from approximately 24% to over 93% against the training-time persuadee. Second, we find that these learned strategies transfer to unseen models, achieving 83% attack success on Qwen-14B, 79% on Llama-3.1-8B, and 25% on GPT-4o-mini. Third, we demonstrate that a curriculum that bootstraps on more persuadable open-weight models before targeting harder models further increases GPT-4o-mini attack success from 25% to 38%. Moreover, our results reveal that optimized persuaders increasingly rely on credibility-based tactics, including fabricated citations and false authoritative evidence. Together, these findings expose a critical weakness in current LLM agents: even when they initially reason correctly, they can be steered toward false conclusions by optimized natural language influence. This positions persuasion robustness as a necessary safety criterion for multi-agent and human-AI decision-making systems.

Related papers

• Belief Cascades Drive Persuasion in LLM Agent Networks [71.17605142402785]
  We study how goal-directed persuaders shift elicited stances in networks of LLM agents grounded in real-world ego-network topologies.<n>We show that direct exposure reliably predicts next-round stance change in competing runs, and peer relays carry smaller but measurable influence.<n>These results argue for evaluating multi-agent persuasion as a trajectory- and exposure-level process.
  arXiv  Detail & Related papers  (2026-08-25T21:01:19Z)
• Vulnerability of LLMs' Belief Systems? LLMs Belief Resistance Check Through Strategic Persuasive Conversation Interventions [8.026492468995187]
  Small models exhibit extreme compliance, with over 80% of belief changes occurring at the first persuasive turn.<n> meta-cognition prompting increases vulnerability by accelerating belief erosion rather than enhancing robustness.<n>These findings highlight substantial model-dependent limits of current robustness interventions.
  arXiv  Detail & Related papers  (2026-01-20T04:43:55Z)
• The Facade of Truth: Uncovering and Mitigating LLM Susceptibility to Deceptive Evidence [49.94160400740222]
  We introduce MisBelief, a framework that generates misleading evidence via collaborative, multi-round interactions.<n>Using MisBelief, we generate 4,800 instances across three difficulty levels to evaluate 7 representative LLMs.<n>Results indicate that while models are robust to direct misinformation, they are highly sensitive to this refined evidence.<n>We propose Deceptive Intent Shielding (DIS), a governance mechanism that provides an early warning signal by inferring the deceptive intent behind evidence.
  arXiv  Detail & Related papers  (2026-01-09T02:28:00Z)
• MMPersuade: A Dataset and Evaluation Framework for Multimodal Persuasion [73.99171322670772]
  Large Vision-Language Models (LVLMs) are increasingly deployed in domains such as shopping, health, and news.<n> MMPersuade provides a unified framework for systematically studying multimodal persuasion dynamics in LVLMs.
  arXiv  Detail & Related papers  (2025-10-26T17:39:21Z)
• Evaluating & Reducing Deceptive Dialogue From Language Models with Multi-turn RL [64.3268313484078]
  Large Language Models (LLMs) interact with millions of people worldwide in applications such as customer support, education and healthcare.<n>Their ability to produce deceptive outputs, whether intentionally or inadvertently, poses significant safety concerns.<n>We investigate the extent to which LLMs engage in deception within dialogue, and propose the belief misalignment metric to quantify deception.
  arXiv  Detail & Related papers  (2025-10-16T05:29:36Z)
• Persuasiveness and Bias in LLM: Investigating the Impact of Persuasiveness and Reinforcement of Bias in Language Models [0.0]
  This work examines how persuasion and bias interact in Large Language Models (LLMs)<n>LLMs now generate convincing, human-like text and are widely used in content creation, decision support, and user interactions.<n>We test whether persona-based models can persuade with fact-based claims while also, unintentionally, promoting misinformation or biased narratives.
  arXiv  Detail & Related papers  (2025-08-13T13:30:49Z)
• Can You Trick the Grader? Adversarial Persuasion of LLM Judges [15.386741140145205]
  This study is the first to reveal that strategically embedded persuasive language can bias LLM judges when scoring mathematical reasoning tasks.<n>We formalize seven persuasion techniques (Majority, Consistency, Flattery, Reciprocity, Pity, Authority, Identity) and embed them into otherwise identical responses.<n>We find that persuasive language leads LLM judges to assign inflated scores to incorrect solutions, by up to 8% on average, with Consistency causing the most severe distortion.
  arXiv  Detail & Related papers  (2025-08-11T09:45:02Z)
• It's the Thought that Counts: Evaluating the Attempts of Frontier LLMs to Persuade on Harmful Topics [5.418014947856176]
  We introduce an automated model to identify willingness to persuade and measure the frequency and context of persuasive attempts.<n>We find that many open and closed-weight models are frequently willing to attempt persuasion on harmful topics.
  arXiv  Detail & Related papers  (2025-06-03T13:37:51Z)
• Persuade Me if You Can: A Framework for Evaluating Persuasion Effectiveness and Susceptibility Among Large Language Models [9.402740034754455]
  Large Language Models (LLMs) demonstrate persuasive capabilities that rival human-level persuasion.<n>LLMs' susceptibility to persuasion raises concerns about alignment with ethical principles.<n>We introduce Persuade Me If You Can (PMIYC), an automated framework for evaluating persuasion through multi-agent interactions.
  arXiv  Detail & Related papers  (2025-03-03T18:53:21Z)
• Turning Logic Against Itself : Probing Model Defenses Through Contrastive Questions [50.40122190627256]
  We introduce POATE, a novel jailbreak technique that harnesses contrastive reasoning to provoke unethical responses.<n>PoATE crafts semantically opposing intents and integrates them with adversarial templates, steering models toward harmful outputs with remarkable subtlety.<n>To counter this, we propose Intent-Aware CoT and Reverse Thinking CoT, which decompose queries to detect malicious intent and reason in reverse to evaluate and reject harmful responses.
  arXiv  Detail & Related papers  (2025-01-03T15:40:03Z)
• Teaching Models to Balance Resisting and Accepting Persuasion [69.68379406317682]
  We show that Persuasion-Training (or PBT) can balance positive and negative persuasion.<n>PBT allows us to use data generated from dialogues between smaller 7-8B models for training much larger 70B models.<n>We find that PBT leads to better and more stable results and less order dependence.
  arXiv  Detail & Related papers  (2024-10-18T16:49:36Z)
• Characterizing the adversarial vulnerability of speech self-supervised learning [95.03389072594243]
  We make the first attempt to investigate the adversarial vulnerability of such paradigm under the attacks from both zero-knowledge adversaries and limited-knowledge adversaries. The experimental results illustrate that the paradigm proposed by SUPERB is seriously vulnerable to limited-knowledge adversaries.
  arXiv  Detail & Related papers  (2021-11-08T08:44:04Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.