Tag
This arXiv paper studies 'mind viruses'—self-propagating ideas that can spread through multi-agent LLM systems—demonstrating with evolutionary algorithms that they can persist across teams and context wipes, while identifying factors that influence spread and proposing mitigation strategies.