Building Resilient Information Ecosystems: Large LLM-Generated Dataset of Persuasion Attacks

2025-11-23

AI Summary

Generative AI rapidly produces persuasive yet misleading content, undermining governmental and corporate credibility; however, existing institutions lack awareness of and defenses against such persuasive strategies. Method: We introduce the first systematic, multi-model persuasive attack dataset—comprising 134,000 samples across GPT-4, Gemma 2, and Llama 3.1—grounded in the 23 persuasion techniques from SemEval 2023. The dataset features both long and short adversarial texts targeting official government and corporate press releases, with quantitative analysis of model preferences across moral foundations (e.g., Care, Authority, Loyalty). Contribution/Results: Our findings uncover structural biases in large language models’ moral resonance, revealing systematic disparities in ethical alignment across models and domains. This work provides an interpretable theoretical framework and empirically validated insights to support organizational “reputation shielding,” advancing information ecosystem resilience from reactive mitigation toward proactive defense.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Machine Learning: Large Multimodal Models (LMMs)
• Application Domains: Humanities & Computational Social Science

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Organization's communication is essential for public trust, but the rise of generative AI models has introduced significant challenges by generating persuasive content that can form competing narratives with official messages from government and commercial organizations at speed and scale. This has left agencies in a reactive position, often unaware of how these models construct their persuasive strategies, making it more difficult to sustain communication effectiveness. In this paper, we introduce a large LLM-generated persuasion attack dataset, which includes 134,136 attacks generated by GPT-4, Gemma 2, and Llama 3.1 on agency news. These attacks span 23 persuasive techniques from SemEval 2023 Task 3, directed toward 972 press releases from ten agencies. The generated attacks come in two mediums, press release statements and social media posts, covering both long-form and short-form communication strategies. We analyzed the moral resonance of these persuasion attacks to understand their attack vectors. GPT-4's attacks mainly focus on Care, with Authority and Loyalty also playing a role. Gemma 2 emphasizes Care and Authority, while Llama 3.1 centers on Loyalty and Care. Analyzing LLM-generated persuasive attacks across models will enable proactive defense, allow to create the reputation armor for organizations, and propel the development of both effective and resilient communications in the information ecosystem.

Problem

• Analyzing persuasion attacks generated by large language models on organizational communications
• Understanding moral resonance and attack vectors in AI-generated persuasive content
• Developing proactive defenses against competing narratives in information ecosystems

Innovation

• Generated large dataset of persuasion attacks using multiple LLMs
• Analyzed moral resonance to understand persuasion attack vectors
• Enabled proactive defense strategies through cross-model attack analysis