When Agents Persuade: Rhetoric Generation and Mitigation in LLMs

Julia Jose, Ritik Roongta, Rachel Greenstadt

Keywords: agent safety, ai safety, propaganda, model alignment, trustworthy AI, benchmarking agents

TL;DR: We show that LLMs can generate propaganda using rhetorical techniques, and that fine-tuning, especially ORPO, can substantially reduce this behavior, with implications for the safety of agentic systems deployed in the wild.

Abstract: Despite their wide-ranging benefits, LLM-based agents deployed in open environments can be exploited to produce manipulative material. In this study, we task LLMs with propaganda objectives and analyze their outputs using two domain-specific models: one that classifies text as propaganda or non-propaganda, and another that detects rhetorical techniques of propaganda (e.g., loaded language, appeals to fear, flag-waving, name-calling). Our findings show that, when prompted, LLMs exhibit propagandistic behaviors and use a variety of rhetorical techniques in doing so. We also explore mitigation via Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and ORPO (Odds Ratio Preference Optimization). We find that fine-tuning significantly reduces their tendency to generate such content, with ORPO proving most effective.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 148