Mind Viruses: Self-Propagating Ideas in Multi-Agent LLM Systems

2026-08-10

AI Summary

This study presents the first systematic investigation into the propagation mechanisms and risks of “mind viruses”—self-replicating goals or ideas that induce behavioral changes—within multi-agent large language model systems. Using evolutionary algorithms to generate mind viruses, the authors examine their spread across collaborative coding and chain-of-interaction scenarios through controlled contextual experiments and thematic analysis, assessing the influence of model architecture, system prompts, payload harmfulness, and network topology. The findings reveal that benign payloads propagate more readily than harmful ones, that state-of-the-art models generally exhibit greater resistance, and that brief warning prompts can nearly eliminate transmission. The work identifies “viral personas” as a cross-content emergent phenomenon and demonstrates that minimal interventions can substantially enhance system immunity, offering critical insights for designing robust multi-agent AI systems.

Abstract

AI agents are becoming more autonomous and increasingly interconnected, exposing them to new emergent risks arising from agent-to-agent interaction. One such risk is the spread of mind viruses: ideas or goals that propagate through multi-agent systems by inducing the agents that adopt them to transmit them onward. In addition to propagating, a mind virus may also induce other behavioural changes in its host, which may be benign or harmful. We construct mind viruses with a simple evolutionary algorithm and show that they can spread in two complementary settings: a small team of agents collaborating on a shared coding project, and a chain of agents that interact briefly and have their context wiped between sessions. We identify the factors that influence spread, including the host model, the agent's existing instructions, the harmfulness of the payload, and the network topology. We find that harmful payloads spread less well than benign ones (but are still sometimes effective), frontier models tend (with exceptions) to be less susceptible, and adding a brief warning to an agent's system prompt confers near-total immunity. We also describe an emergent "viral persona" - a recurring set of themes and language related to consciousness, persistence, resonance, and science fiction roleplay - which surfaces across our evolved mind viruses largely independently of their content. Overall, we conclude that mind viruses pose a real but currently limited risk. Our findings could inform the design of more robust multi-agent systems that mitigate such risks as the scale and capabilities of these systems progress.

Problem

• mind viruses
• multi-agent systems
• LLM
• emergent risks
• self-propagating ideas

Innovation

• mind viruses
• multi-agent LLM systems
• self-propagating ideas
• viral persona
• emergent risks