Metraipolis › archive › dc_cca503f5e474

Learning to Persuade Exposes How Easily LLMs Abandon Correct Beliefs — arXiv:2608.11624

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.11624
License (as archived)
CC0-1.0
Archived
2026-10-19T05:16:01.124Z
SHA-256 of the archived text
d59a80dff37bc3d4c8d0bdc2f48d39214421cad180a92d353e4b0dca55e5872b
Stable URL of this page
https://metraipolis.com/documento/dc_cca503f5e474

How to cite

Learning to Persuade Exposes How Easily LLMs Abandon Correct Beliefs — arXiv:2608.11624 — arXiv (en), https://arxiv.org/abs/2608.11624 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cca503f5e474 (Metraipolis archive dc_cca503f5e474, sha256 d59a80dff37bc3d4c8d0bdc2f48d39214421cad180a92d353e4b0dca55e5872b)
BibTeX
@misc{dccca503f5e474,
  title = {Learning to Persuade Exposes How Easily LLMs Abandon Correct Beliefs — arXiv:2608.11624},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.11624},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cca503f5e474; sha256 d59a80dff37bc3d4c8d0bdc2f48d39214421cad180a92d353e4b0dca55e5872b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.11624. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

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.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-13 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.11624 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-19T05:08:46.707Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-13?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).