|Jul 12
FIFA World Cup 2026
Watch Live →
Science

Adversarial Social Epistemology: Trust and Misinformation in Human-LLM Interactions

Mihnea C. Moldoveanu and Joel A.C. Baum discuss trust and misinformation in their new paper on adversarial social epistemology.

By Feed and Figures Editorial Team1 min readSource: arXiv AI
Research paper titled Adversarial Social Epistemology for Assemblies of Humans and Large Language Models on a digital screen

On July 8, 2026, Mihnea C. Moldoveanu and Joel A.C. Baum presented a paper titled Adversarial Social Epistemology for Assemblies of Humans and Large Language Models. This work outlines a framework aimed at understanding the complexities of trust and misinformation within interactive communication environments shaped by both human and AI agents.

Understanding Adversarial Social Epistemology

The authors propose a model of Adversarial Social Epistemology (ASE) that examines how public assertions in communication are influenced by chains of testimony, inference, and trust. They argue that in modern communication landscapes, agents may intentionally distort or fabricate information for various personal gains.

This analysis moves beyond traditional concepts like epistemic bubbles and echo chambers, highlighting the need for a deeper understanding of how these dynamics operate in real-world scenarios. Moldoveanu and Baum emphasize the importance of recognizing the mechanisms that lead to trust breaches among communicative agents.

Mechanisms of Trust Breach in Communication

The ASE framework identifies specific mechanisms that can undermine trust in public assertions. These include:

  • Distortion of Information: Agents may omit or strategically under-specify details to mislead others.
  • Fabrication: Creating false information to serve personal interests.
  • Rhetorical Manipulation: Using persuasive language to influence perceptions without factual basis.

By analyzing these behaviors, the authors aim to develop a comprehensive understanding of how trust can be systematically eroded in communication networks.

Auditing Trust Breaches and Restoring Integrity

Moldoveanu and Baum propose methodologies for auditing and redressing breaches of trust in public communications. They advocate for enhanced auditability of inferential chains, which can help in identifying and mitigating misinformation spread.

The incorporation of epistemic networks and an inferentialist semantics model provides a robust framework for interpreting assertions and evaluating their trustworthiness. This approach aims to empower individuals and institutions to better navigate the complexities of information in the digital age.

🤖 This article was rewritten by Feed and Figures' editorial AI from a report originally published by arXiv AI. Facts and quotes are preserved from the original; the rewrite focuses on clarity and structure. For the unedited original, see the source link below.

#Mihnea C. Moldoveanu
#Joel A.C. Baum
#artificial intelligence
#social networks
#epistemology

Related stories